diff options
Diffstat (limited to 'clang/test')
169 files changed, 69015 insertions, 0 deletions
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfadd.c new file mode 100644 index 0000000..d7734e0 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfadd.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfclass.c new file mode 100644 index 0000000..68814f4 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfclass.c @@ -0,0 +1,134 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf4_u16mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf2_u16mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m1_u16m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m2_u16m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m4_u16m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m8_u16m8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_m(vbool64_t vm, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16mf4_u16mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16mf2_u16mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m1_u16m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m2_u16m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m4_u16m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m8_u16m8_m(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmacc.c new file mode 100644 index 0000000..616455d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmacc.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmadd.c new file mode 100644 index 0000000..eec662a --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmadd.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmax.c new file mode 100644 index 0000000..dfdeb4e --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmax.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmerge.c new file mode 100644 index 0000000..96221c5 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmerge.c @@ -0,0 +1,69 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmerge_vfm_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16mf4(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmerge_vfm_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16mf2(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmerge_vfm_bf16m1(vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m1(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmerge_vfm_bf16m2(vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m2(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmerge_vfm_bf16m4(vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m4(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmerge_vfm_bf16m8(vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m8(op1, op2, mask, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmin.c new file mode 100644 index 0000000..8f8d82ba --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmin.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsac.c new file mode 100644 index 0000000..f4644df --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsac.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsub.c new file mode 100644 index 0000000..07053afa --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmul.c new file mode 100644 index 0000000..88fb329 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmul.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmv.c new file mode 100644 index 0000000..d80ec3d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmv.c @@ -0,0 +1,189 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_v_f_bf16mf4( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16.i64(<vscale x 1 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmv_v_f_bf16mf4(__bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16mf4(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_v_f_bf16mf2( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16.i64(<vscale x 2 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmv_v_f_bf16mf2(__bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16mf2(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_v_f_bf16m1( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16.i64(<vscale x 4 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmv_v_f_bf16m1(__bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m1(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_v_f_bf16m2( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16.i64(<vscale x 8 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmv_v_f_bf16m2(__bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m2(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_v_f_bf16m4( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16.i64(<vscale x 16 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmv_v_f_bf16m4(__bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m4(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_v_f_bf16m8( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16.i64(<vscale x 32 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmv_v_f_bf16m8(__bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m8(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf4_bf16( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv1bf16(<vscale x 1 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16mf4_bf16(vbfloat16mf4_t src) { + return __riscv_vfmv_f_s_bf16mf4_bf16(src); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_s_f_bf16mf4( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16.i64(<vscale x 1 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmv_s_f_bf16mf4(__bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16mf4(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf2_bf16( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv2bf16(<vscale x 2 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16mf2_bf16(vbfloat16mf2_t src) { + return __riscv_vfmv_f_s_bf16mf2_bf16(src); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_s_f_bf16mf2( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16.i64(<vscale x 2 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmv_s_f_bf16mf2(__bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16mf2(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m1_bf16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv4bf16(<vscale x 4 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m1_bf16(vbfloat16m1_t src) { + return __riscv_vfmv_f_s_bf16m1_bf16(src); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_s_f_bf16m1( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16.i64(<vscale x 4 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmv_s_f_bf16m1(__bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m1(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m2_bf16( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv8bf16(<vscale x 8 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m2_bf16(vbfloat16m2_t src) { + return __riscv_vfmv_f_s_bf16m2_bf16(src); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_s_f_bf16m2( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16.i64(<vscale x 8 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmv_s_f_bf16m2(__bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m2(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m4_bf16( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv16bf16(<vscale x 16 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m4_bf16(vbfloat16m4_t src) { + return __riscv_vfmv_f_s_bf16m4_bf16(src); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_s_f_bf16m4( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16.i64(<vscale x 16 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmv_s_f_bf16m4(__bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m4(src, vl); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m8_bf16( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv32bf16(<vscale x 32 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m8_bf16(vbfloat16m8_t src) { + return __riscv_vfmv_f_s_bf16m8_bf16(src); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_s_f_bf16m8( +// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16.i64(<vscale x 32 x bfloat> poison, bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmv_s_f_bf16m8(__bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m8(src, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt.c new file mode 100644 index 0000000..a5afab9 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt.c @@ -0,0 +1,724 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm(vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm(vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm(vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm(vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm(vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_rm(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(vbool32_t vm, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(vbool32_t vm, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(vbool16_t vm, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_m(vbool64_t vm, vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_m(vbool32_t vm, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_m(vbool16_t vm, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_m(vbool8_t vm, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_m(vbool4_t vm, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rod.c new file mode 100644 index 0000000..70c377b --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rod.c @@ -0,0 +1,113 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m4_m(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rtz.c new file mode 100644 index 0000000..854e986 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rtz.c @@ -0,0 +1,267 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmacc.c new file mode 100644 index 0000000..1848488 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmacc.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmadd.c new file mode 100644 index 0000000..e519e5a --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmadd.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsac.c new file mode 100644 index 0000000..47e1f44 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsac.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsub.c new file mode 100644 index 0000000..4b55b64 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m1(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m2(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m4(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m8(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m1_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m1_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m2_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m2_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m4_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m4_m(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m8_m(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m8_m(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrec7.c new file mode 100644 index 0000000..1ffee73 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrec7.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf4(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf2(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1(vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m1(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2(vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m2(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4(vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m4(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8(vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m8(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf4_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf2_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m1_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m2_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m4_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m8_m(mask, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsqrt7.c new file mode 100644 index 0000000..964c486 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsqrt7.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf4(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf2(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1(vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m1(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2(vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m2(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4(vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m4(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8(vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m8(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf4_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf2_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m1_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m2_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m4_m(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m8_m(mask, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsub.c new file mode 100644 index 0000000..c7c3869e --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsub.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnj.c new file mode 100644 index 0000000..778b8b83 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnj.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjn.c new file mode 100644 index 0000000..7de3089 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjn.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjx.c new file mode 100644 index 0000000..5fa285cc --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjx.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1down.c new file mode 100644 index 0000000..b94d26b --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1down.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf4(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf2(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m1(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m2(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m4(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m8(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf4_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf2_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m1_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m2_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m4_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m8_m(mask, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1up.c new file mode 100644 index 0000000..06e8b49 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1up.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf4(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf2(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m1(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m2(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m4(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m8(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf4_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf2_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m1_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m2_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m4_m(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m8_m(mask, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsub.c new file mode 100644 index 0000000..2423b0b --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m1(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m1_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m8_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwadd.c new file mode 100644 index 0000000..24d34f4 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwadd.c @@ -0,0 +1,899 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwcvt.c new file mode 100644 index 0000000..fb3e003 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwcvt.c @@ -0,0 +1,366 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4(vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2(vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1(vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2(vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4(vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8(vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4(vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2(vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1(vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2(vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4(vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8(vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m1_f32m2(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m2_f32m4(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m4_f32m8(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_m(vbool64_t vm, vint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_m(vbool32_t vm, vint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_m(vbool16_t vm, vint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_m(vbool8_t vm, vint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_m(vbool4_t vm, vint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_m(vbool2_t vm, vint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m8_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_m(vbool64_t vm, vuint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_m(vbool32_t vm, vuint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_m(vbool16_t vm, vuint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_m(vbool8_t vm, vuint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_m(vbool4_t vm, vuint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_m(vbool2_t vm, vuint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m8_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_m(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_m(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmacc.c new file mode 100644 index 0000000..be09003 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmacc.c @@ -0,0 +1,486 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmsac.c new file mode 100644 index 0000000..7490813 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmsac.c @@ -0,0 +1,486 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmul.c new file mode 100644 index 0000000..6783ba4 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmul.c @@ -0,0 +1,455 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmacc.c new file mode 100644 index 0000000..6127a94 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmacc.c @@ -0,0 +1,494 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmsac.c new file mode 100644 index 0000000..f37dd31 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmsac.c @@ -0,0 +1,494 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwsub.c new file mode 100644 index 0000000..510ff91 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwsub.c @@ -0,0 +1,899 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_m(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_m(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfeq.c new file mode 100644 index 0000000..669d042 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfeq.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m8_b2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m8_b2_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfge.c new file mode 100644 index 0000000..b169efd --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfge.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m8_b2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m8_b2_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfgt.c new file mode 100644 index 0000000..9aea7d2 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfgt.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m8_b2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m8_b2_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfle.c new file mode 100644 index 0000000..40f0c27 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfle.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m8_b2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m8_b2_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmflt.c new file mode 100644 index 0000000..f64eee3 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmflt.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m8_b2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m8_b2_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfne.c new file mode 100644 index 0000000..809ea56 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfne.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16mf4_b64(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16mf2_b32(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m1_b16(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m2_b8(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m4_b4(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m8_b2(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16mf4_b64_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16mf2_b32_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m1_b16_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m2_b8_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m4_b4_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m8_b2_m(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m8_b2_m(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfadd.c new file mode 100644 index 0000000..9d6b071c --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfadd.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfclass.c new file mode 100644 index 0000000..2760f85 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfclass.c @@ -0,0 +1,134 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_m(vbool64_t vm, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfclass(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfclass(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfclass(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfclass(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfclass(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfclass(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmacc.c new file mode 100644 index 0000000..ae3f1f2 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmacc.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmadd.c new file mode 100644 index 0000000..db2184c --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmadd.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmax.c new file mode 100644 index 0000000..66497bf --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmax.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmerge.c new file mode 100644 index 0000000..1dc290b --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmerge.c @@ -0,0 +1,69 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmerge_vfm_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) { + return __riscv_vfmerge(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmerge_vfm_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) { + return __riscv_vfmerge(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmerge_vfm_bf16m1(vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) { + return __riscv_vfmerge(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmerge_vfm_bf16m2(vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) { + return __riscv_vfmerge(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmerge_vfm_bf16m4(vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) { + return __riscv_vfmerge(op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmerge_vfm_bf16m8(vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) { + return __riscv_vfmerge(op1, op2, mask, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmin.c new file mode 100644 index 0000000..1564d11 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmin.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsac.c new file mode 100644 index 0000000..0384e7d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsac.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsub.c new file mode 100644 index 0000000..306f189 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmul.c new file mode 100644 index 0000000..fffd83a --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmul.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmv.c new file mode 100644 index 0000000..f85378f --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmv.c @@ -0,0 +1,69 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf4_bf16( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv1bf16(<vscale x 1 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16mf4_bf16(vbfloat16mf4_t src) { + return __riscv_vfmv_f(src); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf2_bf16( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv2bf16(<vscale x 2 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16mf2_bf16(vbfloat16mf2_t src) { + return __riscv_vfmv_f(src); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m1_bf16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv4bf16(<vscale x 4 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m1_bf16(vbfloat16m1_t src) { + return __riscv_vfmv_f(src); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m2_bf16( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv8bf16(<vscale x 8 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m2_bf16(vbfloat16m2_t src) { + return __riscv_vfmv_f(src); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m4_bf16( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv16bf16(<vscale x 16 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m4_bf16(vbfloat16m4_t src) { + return __riscv_vfmv_f(src); +} + +// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m8_bf16( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv32bf16(<vscale x 32 x bfloat> [[SRC]]) +// CHECK-RV64-NEXT: ret bfloat [[TMP0]] +// +__bf16 test_vfmv_f_s_bf16m8_bf16(vbfloat16m8_t src) { + return __riscv_vfmv_f(src); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt.c new file mode 100644 index 0000000..fb635d6 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt.c @@ -0,0 +1,724 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm(vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm(vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm(vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm(vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm(vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(vbool32_t vm, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(vbool32_t vm, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(vbool16_t vm, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_m(vbool64_t vm, vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_m(vbool32_t vm, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_m(vbool16_t vm, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_m(vbool8_t vm, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_m(vbool4_t vm, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rod.c new file mode 100644 index 0000000..1ad856d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rod.c @@ -0,0 +1,113 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rtz.c new file mode 100644 index 0000000..12d0893 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rtz.c @@ -0,0 +1,267 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmacc.c new file mode 100644 index 0000000..6f7928b --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmacc.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmadd.c new file mode 100644 index 0000000..97d2070 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmadd.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsac.c new file mode 100644 index 0000000..404b4f8 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsac.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsub.c new file mode 100644 index 0000000..3a520dd --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrec7.c new file mode 100644 index 0000000..462b6ac --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrec7.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1(vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2(vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4(vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8(vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7(mask, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsqrt7.c new file mode 100644 index 0000000..051fde7 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsqrt7.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1(vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2(vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4(vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8(vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7(op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7(mask, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7(mask, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsub.c new file mode 100644 index 0000000..0494182 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsub.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnj.c new file mode 100644 index 0000000..615dedd --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnj.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjn.c new file mode 100644 index 0000000..a895e5f --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjn.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjx.c new file mode 100644 index 0000000..0187516 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjx.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1down.c new file mode 100644 index 0000000..4a76894 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1down.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down(mask, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1up.c new file mode 100644 index 0000000..f9f2dc0 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1up.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(mask, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up(mask, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsub.c new file mode 100644 index 0000000..ebcf6fa --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwadd.c new file mode 100644 index 0000000..124e7fb --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwadd.c @@ -0,0 +1,893 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwcvt.c new file mode 100644 index 0000000..0399a63 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwcvt.c @@ -0,0 +1,366 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4(vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2(vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1(vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2(vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4(vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8(vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4(vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2(vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1(vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2(vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4(vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8(vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2(vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1(vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2(vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4(vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8(vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f(vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_m(vbool64_t vm, vint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_m(vbool32_t vm, vint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_m(vbool16_t vm, vint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_m(vbool8_t vm, vint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_m(vbool4_t vm, vint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_m(vbool2_t vm, vint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_m(vbool64_t vm, vuint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_m(vbool32_t vm, vuint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_m(vbool16_t vm, vuint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_m(vbool8_t vm, vuint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_m(vbool4_t vm, vuint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_m(vbool2_t vm, vuint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_m(vbool64_t vm, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f(vm, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f(vm, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmacc.c new file mode 100644 index 0000000..2eb7fc8 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmacc.c @@ -0,0 +1,474 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmsac.c new file mode 100644 index 0000000..28f5076 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmsac.c @@ -0,0 +1,474 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmul.c new file mode 100644 index 0000000..8de49fa --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmul.c @@ -0,0 +1,451 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwmul(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwmul(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwmul(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmacc.c new file mode 100644 index 0000000..7836931 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmacc.c @@ -0,0 +1,480 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmsac.c new file mode 100644 index 0000000..ca936af --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmsac.c @@ -0,0 +1,480 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwsub.c new file mode 100644 index 0000000..2e22e22 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwsub.c @@ -0,0 +1,893 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfeq.c new file mode 100644 index 0000000..29881c9 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfeq.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfge.c new file mode 100644 index 0000000..b8083c5e --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfge.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfgt.c new file mode 100644 index 0000000..b8749b3 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfgt.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfle.c new file mode 100644 index 0000000..724608c --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfle.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmflt.c new file mode 100644 index 0000000..1b0b898 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmflt.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfne.c new file mode 100644 index 0000000..672c150 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfne.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_m( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_m( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_m( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_m( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_m( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_m( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne(mask, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfadd.c new file mode 100644 index 0000000..6d55279 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfadd.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfclass.c new file mode 100644 index 0000000..8e6946d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfclass.c @@ -0,0 +1,272 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tu(vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf4_u16mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tu(vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf2_u16mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_tu(vuint16m1_t vd, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m1_u16m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_tu(vuint16m2_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m2_u16m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_tu(vuint16m4_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m4_u16m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_tu(vuint16m8_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfclass_v_bf16m8_u16m8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tum(vbool64_t vm, vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf4_u16mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tum(vbool32_t vm, vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf2_u16mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_tum(vbool16_t vm, vuint16m1_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m1_u16m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_tum(vbool8_t vm, vuint16m2_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m2_u16m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_tum(vbool4_t vm, vuint16m4_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m4_u16m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_tum(vbool2_t vm, vuint16m8_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m8_u16m8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tumu(vbool64_t vm, vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf4_u16mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tumu(vbool32_t vm, vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf2_u16mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_tumu(vbool16_t vm, vuint16m1_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m1_u16m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_tumu(vbool8_t vm, vuint16m2_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m2_u16m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_tumu(vbool4_t vm, vuint16m4_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m4_u16m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_tumu(vbool2_t vm, vuint16m8_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m8_u16m8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_mu(vbool64_t vm, vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf4_u16mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_mu(vbool32_t vm, vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16mf2_u16mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_mu(vbool16_t vm, vuint16m1_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m1_u16m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_mu(vbool8_t vm, vuint16m2_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m2_u16m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_mu(vbool4_t vm, vuint16m4_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m4_u16m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_mu(vbool2_t vm, vuint16m8_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass_v_bf16m8_u16m8_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmacc.c new file mode 100644 index 0000000..2d4e481 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmacc.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmadd.c new file mode 100644 index 0000000..511e073 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmadd.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmax.c new file mode 100644 index 0000000..f3698d4 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmax.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmerge.c new file mode 100644 index 0000000..bcaf2cb --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmerge.c @@ -0,0 +1,69 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmerge_vfm_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16mf4_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmerge_vfm_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16mf2_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmerge_vfm_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m1_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmerge_vfm_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m2_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmerge_vfm_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m4_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmerge_vfm_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) { + return __riscv_vfmerge_vfm_bf16m8_tu(maskedoff, op1, op2, mask, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmin.c new file mode 100644 index 0000000..911f879 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmin.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsac.c new file mode 100644 index 0000000..9575ad3 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsac.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsub.c new file mode 100644 index 0000000..8e382f71 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsub.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmul.c new file mode 100644 index 0000000..716f056 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmul.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmv.c new file mode 100644 index 0000000..069ee6a --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmv.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_v_f_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmv_v_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16mf4_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_v_f_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmv_v_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16mf2_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_v_f_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmv_v_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m1_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_v_f_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmv_v_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m2_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_v_f_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmv_v_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m4_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_v_f_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmv_v_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_f_bf16m8_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_s_f_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmv_s_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16mf4_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_s_f_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmv_s_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16mf2_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_s_f_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmv_s_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m1_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_s_f_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmv_s_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m2_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_s_f_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmv_s_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m4_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_s_f_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmv_s_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_f_bf16m8_tu(maskedoff, src, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt.c new file mode 100644 index 0000000..36d4fc3 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt.c @@ -0,0 +1,1577 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tu(vbfloat16m1_t vd, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tu(vbfloat16m2_t vd, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tu(vbfloat16m4_t vd, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(vint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(vint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(vint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tu(vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tu(vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tu(vbfloat16m1_t vd, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tu(vbfloat16m2_t vd, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tu(vbfloat16m4_t vd, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(vm, vd, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(vm, vd, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tum(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tum(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tum(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tum(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tum(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(vm, vd, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(vm, vd, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(vm, vd, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(vm, vd, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(vbool16_t vm, + vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(vm, vd, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tumu(vbool64_t vm, + vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tumu(vbool32_t vm, + vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tumu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tumu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tumu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_mu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_mu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16mf2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_mu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m1_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_mu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_mu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_f_w_bf16m4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rod.c new file mode 100644 index 0000000..8406684 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rod.c @@ -0,0 +1,233 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tu(vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tu(vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tu(vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tum(vbool64_t vm, + vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tum(vbool32_t vm, + vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tumu(vbool64_t vm, + vbfloat16mf4_t vd, + vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tumu(vbool32_t vm, + vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_f_w_bf16m4_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rtz.c new file mode 100644 index 0000000..4644eff --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rtz.c @@ -0,0 +1,572 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm, + vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm, + vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmacc.c new file mode 100644 index 0000000..93fd6ba --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmacc.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmadd.c new file mode 100644 index 0000000..d7e6b82 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmadd.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsac.c new file mode 100644 index 0000000..e0c289d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsac.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsub.c new file mode 100644 index 0000000..05ccda3 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsub.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m1_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m2_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m4_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m8_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m1_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m1_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m2_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m2_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m4_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m4_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m8_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m8_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m1_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m1_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m2_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m2_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m4_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m4_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vv_bf16m8_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_vf_bf16m8_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrec7.c new file mode 100644 index 0000000..3123692 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrec7.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf4_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf2_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m1_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m2_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m4_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m8_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf4_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf2_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m1_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m2_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m4_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m8_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf4_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf2_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m1_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m2_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m4_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m8_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf4_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16mf2_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m1_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m2_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m4_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_v_bf16m8_mu(mask, maskedoff, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsqrt7.c new file mode 100644 index 0000000..8436f0e --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsqrt7.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf4_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf2_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m1_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m2_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m4_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m8_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf4_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf2_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m1_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m2_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m4_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m8_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf4_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf2_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m1_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m2_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m4_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m8_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf4_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16mf2_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m1_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m2_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m4_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_v_bf16m8_mu(mask, maskedoff, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsub.c new file mode 100644 index 0000000..7dd2bb6 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnj.c new file mode 100644 index 0000000..b39a0be --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnj.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjn.c new file mode 100644 index 0000000..7542e78 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjn.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjx.c new file mode 100644 index 0000000..104149e --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjx.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1down.c new file mode 100644 index 0000000..228dc1cd --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1down.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf4_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf2_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m1_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m2_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m4_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m8_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf4_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf2_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m1_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m2_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m4_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m8_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf4_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf2_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m1_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m2_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m4_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m8_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf4_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16mf2_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m1_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m2_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m4_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_vf_bf16m8_mu(mask, maskedoff, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1up.c new file mode 100644 index 0000000..9e6ff2b --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1up.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf4_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf2_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m1_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m2_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m4_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m8_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf4_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf2_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m1_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m2_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m4_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m8_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf4_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf2_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m1_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m2_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m4_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m8_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf4_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16mf2_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m1_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m2_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m4_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_vf_bf16m8_mu(mask, maskedoff, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsub.c new file mode 100644 index 0000000..b6fd94e --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsub.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m1_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m2_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m4_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m8_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwadd.c new file mode 100644 index 0000000..4bee376 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwadd.c @@ -0,0 +1,2007 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32mf2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m1_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m1_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m2_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_bf16m4_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_bf16_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwcvt.c new file mode 100644 index 0000000..9151319 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwcvt.c @@ -0,0 +1,765 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tu(vbfloat16mf4_t vd, vint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tu(vbfloat16mf2_t vd, vint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tu(vbfloat16m1_t vd, vint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tu(vbfloat16m2_t vd, vint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tu(vbfloat16m4_t vd, vint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tu(vbfloat16m8_t vd, vint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tu(vbfloat16mf4_t vd, vuint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tu(vbfloat16mf2_t vd, vuint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tu(vbfloat16m1_t vd, vuint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tu(vbfloat16m2_t vd, vuint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tu(vbfloat16m4_t vd, vuint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tu(vbfloat16m8_t vd, vuint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd, + vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd, + vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd, + vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd, + vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd, + vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd, + vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd, + vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd, + vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd, + vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd, + vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd, + vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd, + vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd, + vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_x_v_bf16m8_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd, + vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_xu_v_bf16m8_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmacc.c new file mode 100644 index 0000000..f67b100 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmacc.c @@ -0,0 +1,1017 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmsac.c new file mode 100644 index 0000000..6d78c74 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmsac.c @@ -0,0 +1,1017 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmul.c new file mode 100644 index 0000000..9fcfe81 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmul.c @@ -0,0 +1,1015 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m1_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m2_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_vv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_vf_bf16m4_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmacc.c new file mode 100644 index 0000000..73cc822 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmacc.c @@ -0,0 +1,1034 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmsac.c new file mode 100644 index 0000000..6133230 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmsac.c @@ -0,0 +1,1034 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2, + __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwsub.c new file mode 100644 index 0000000..9d9b0b0 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwsub.c @@ -0,0 +1,2007 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_rm_tum(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32mf2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs2, rs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1, + __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m1_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m1_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m2_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_bf16m4_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, + vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_bf16_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, + vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfeq.c new file mode 100644 index 0000000..b96aae5 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfeq.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfeq_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfge.c new file mode 100644 index 0000000..47d0427 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfge.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfge_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfgt.c new file mode 100644 index 0000000..0a0ead2 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfgt.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfgt_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfle.c new file mode 100644 index 0000000..27ddefe --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfle.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfle_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmflt.c new file mode 100644 index 0000000..d5f4f77 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmflt.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmflt_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfne.c new file mode 100644 index 0000000..c2df947 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfne.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfne_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfadd.c new file mode 100644 index 0000000..2bd3b39 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfadd.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfclass.c new file mode 100644 index 0000000..e2a993a --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfclass.c @@ -0,0 +1,272 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tu(vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tu(vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_tu(vuint16m1_t vd, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfclass_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_tu(vuint16m2_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfclass_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_tu(vuint16m4_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfclass_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_tu(vuint16m8_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfclass_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tum(vbool64_t vm, vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tum(vbool32_t vm, vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_tum(vbool16_t vm, vuint16m1_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_tum(vbool8_t vm, vuint16m2_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_tum(vbool4_t vm, vuint16m4_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_tum(vbool2_t vm, vuint16m8_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tumu(vbool64_t vm, vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tumu(vbool32_t vm, vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_tumu(vbool16_t vm, vuint16m1_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_tumu(vbool8_t vm, vuint16m2_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_tumu(vbool4_t vm, vuint16m4_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_tumu(vbool2_t vm, vuint16m8_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i16> [[TMP0]] +// +vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_mu(vbool64_t vm, vuint16mf4_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfclass_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i16> [[TMP0]] +// +vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_mu(vbool32_t vm, vuint16mf2_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfclass_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i16> [[TMP0]] +// +vuint16m1_t test_vfclass_v_bf16m1_u16m1_mu(vbool16_t vm, vuint16m1_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfclass_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i16> [[TMP0]] +// +vuint16m2_t test_vfclass_v_bf16m2_u16m2_mu(vbool8_t vm, vuint16m2_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfclass_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i16> [[TMP0]] +// +vuint16m4_t test_vfclass_v_bf16m4_u16m4_mu(vbool4_t vm, vuint16m4_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfclass_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i16> [[TMP0]] +// +vuint16m8_t test_vfclass_v_bf16m8_u16m8_mu(vbool2_t vm, vuint16m8_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfclass_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmacc.c new file mode 100644 index 0000000..eb74271 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmacc.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmadd.c new file mode 100644 index 0000000..68d490d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmadd.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmax.c new file mode 100644 index 0000000..5f682e8 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmax.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmax_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmax_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmax_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmax_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmax_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmax_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmerge.c new file mode 100644 index 0000000..9593ad5 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmerge.c @@ -0,0 +1,69 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmerge_vfm_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) { + return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmerge_vfm_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) { + return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmerge_vfm_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) { + return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmerge_vfm_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) { + return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmerge_vfm_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) { + return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmerge_vfm_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) { + return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmin.c new file mode 100644 index 0000000..f3ef3c3 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmin.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmin_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmin_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmin_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmin_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmin_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmin_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsac.c new file mode 100644 index 0000000..0587c57 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsac.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsub.c new file mode 100644 index 0000000..2ad26f8 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsub.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmul.c new file mode 100644 index 0000000..d1e726a9 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmul.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmul_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmul_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmul_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmul_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmul_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmul_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmv.c new file mode 100644 index 0000000..9fd1ffc --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmv.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_v_f_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmv_v_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_v_f_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmv_v_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_v_f_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmv_v_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_v_f_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmv_v_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_v_f_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmv_v_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_v_f_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmv_v_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_v_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_s_f_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfmv_s_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_s_f_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfmv_s_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_s_f_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfmv_s_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_s_f_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfmv_s_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_s_f_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfmv_s_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_tu(maskedoff, src, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_s_f_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfmv_s_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) { + return __riscv_vfmv_s_tu(maskedoff, src, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt.c new file mode 100644 index 0000000..c6cd0a5 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt.c @@ -0,0 +1,1539 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd, vfloat32m1_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tu(vbfloat16m1_t vd, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tu(vbfloat16m2_t vd, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tu(vbfloat16m4_t vd, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(vint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(vint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(vint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tu(vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tu(vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tu(vbfloat16m1_t vd, vfloat32m2_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tu(vbfloat16m2_t vd, vfloat32m4_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tu(vbfloat16m4_t vd, vfloat32m8_t vs2, + size_t vl) { + return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tum(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tum(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tum(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tum(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tum(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(vbool16_t vm, + vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tumu(vbool64_t vm, + vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tumu(vbool32_t vm, + vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tumu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tumu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tumu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(vbool64_t vm, vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(vbool32_t vm, vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_mu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_mu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_mu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_mu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_mu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rod.c new file mode 100644 index 0000000..0745633 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rod.c @@ -0,0 +1,233 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tu(vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tu(vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tu(vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tum(vbool64_t vm, + vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tum(vbool32_t vm, + vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tumu(vbool64_t vm, + vbfloat16mf4_t vd, + vfloat32mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tumu(vbool32_t vm, + vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vfloat32mf2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vfloat32m1_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vfloat32m2_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vfloat32m4_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vfloat32m8_t vs2, size_t vl) { + return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rtz.c new file mode 100644 index 0000000..b906c5f --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rtz.c @@ -0,0 +1,572 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu( +// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu( +// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu( +// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu( +// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu( +// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu( +// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm, + vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm, + vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x i8> [[TMP0]] +// +vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm, + vuint8mf8_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x i8> [[TMP0]] +// +vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm, + vuint8mf4_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x i8> [[TMP0]] +// +vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x i8> [[TMP0]] +// +vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x i8> [[TMP0]] +// +vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x i8> [[TMP0]] +// +vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd, + vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmacc.c new file mode 100644 index 0000000..cc487b4 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmacc.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmadd.c new file mode 100644 index 0000000..f9c348b --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmadd.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsac.c new file mode 100644 index 0000000..83d35e8 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsac.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsub.c new file mode 100644 index 0000000..f5282a1 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsub.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_tu(vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfnmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfnmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfnmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfnmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) { + return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrec7.c new file mode 100644 index 0000000..f8e5a33 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrec7.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrec7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrec7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrec7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrec7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrec7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrec7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrec7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrec7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrec7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrec7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrec7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrec7_mu(mask, maskedoff, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsqrt7.c new file mode 100644 index 0000000..7c6c926 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsqrt7.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_tu(maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) { + return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) { + return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsqrt7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) { + return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsqrt7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) { + return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsqrt7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) { + return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsqrt7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) { + return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsub.c new file mode 100644 index 0000000..c09caeb --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsub.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfrsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfrsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfrsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfrsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfrsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfrsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnj.c new file mode 100644 index 0000000..c1f69932 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnj.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnj_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnj_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnj_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnj_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjn.c new file mode 100644 index 0000000..1b799d8 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjn.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjn_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjn_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjn_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjn_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjx.c new file mode 100644 index 0000000..9c5f2af --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjx.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsgnjx_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsgnjx_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsgnjx_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsgnjx_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1down.c new file mode 100644 index 0000000..691302e --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1down.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1down_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1down_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1down_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1down_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1up.c new file mode 100644 index 0000000..1238d22 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1up.c @@ -0,0 +1,249 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tu(maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfslide1up_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfslide1up_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfslide1up_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfslide1up_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) { + return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsub.c new file mode 100644 index 0000000..ea4f8f0 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsub.c @@ -0,0 +1,489 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tu(maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwadd.c new file mode 100644 index 0000000..e5b7b8d --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwadd.c @@ -0,0 +1,1932 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwcvt.c new file mode 100644 index 0000000..7300104 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwcvt.c @@ -0,0 +1,765 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tu(vbfloat16mf4_t vd, vint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tu(vbfloat16mf2_t vd, vint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tu(vbfloat16m1_t vd, vint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tu(vbfloat16m2_t vd, vint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tu(vbfloat16m4_t vd, vint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tu(vbfloat16m8_t vd, vint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tu( +// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tu(vbfloat16mf4_t vd, vuint8mf8_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tu( +// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tu(vbfloat16mf2_t vd, vuint8mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tu( +// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tu(vbfloat16m1_t vd, vuint8mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tu( +// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tu(vbfloat16m2_t vd, vuint8m1_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tu( +// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tu(vbfloat16m4_t vd, vuint8m2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tu( +// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tu(vbfloat16m8_t vd, vuint8m4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tu(vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd, + vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd, + vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd, + vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd, + vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd, + vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd, + vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd, + vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd, + vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tum( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd, + vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd, + vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd, + vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd, + vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd, + vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd, + vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd, + vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd, + vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd, + vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tumu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd, + vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd, + vint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x bfloat> [[TMP0]] +// +vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd, + vuint8mf8_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x bfloat> [[TMP0]] +// +vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd, + vuint8mf4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x bfloat> [[TMP0]] +// +vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd, + vuint8mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x bfloat> [[TMP0]] +// +vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd, + vuint8m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x bfloat> [[TMP0]] +// +vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd, + vuint8m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 32 x bfloat> [[TMP0]] +// +vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd, + vuint8m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmacc.c new file mode 100644 index 0000000..b05f8802 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmacc.c @@ -0,0 +1,977 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmsac.c new file mode 100644 index 0000000..93721f6 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmsac.c @@ -0,0 +1,977 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmul.c new file mode 100644 index 0000000..4a2b5e3 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmul.c @@ -0,0 +1,975 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmacc.c new file mode 100644 index 0000000..57e43344 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmacc.c @@ -0,0 +1,994 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmsac.c new file mode 100644 index 0000000..42da060 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmsac.c @@ -0,0 +1,994 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, + vfloat32mf2_t vd, __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + __bf16 vs1, vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + __bf16 vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, + vfloat32m1_t vd, __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, + vfloat32m2_t vd, __bf16 vs1, + vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, + vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, + vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs1, + vbfloat16mf4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu( + vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + __bf16 vs1, + vbfloat16mf2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs1, + vbfloat16m1_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + __bf16 vs1, vbfloat16m1_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs1, + vbfloat16m2_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + __bf16 vs1, vbfloat16m2_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs1, + vbfloat16m4_t vs2, size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + __bf16 vs1, vbfloat16m4_t vs2, + size_t vl) { + return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwsub.c new file mode 100644 index 0000000..1378bc9 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwsub.c @@ -0,0 +1,1932 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vbfloat16mf4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, vbfloat16m1_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, vbfloat16m2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, vbfloat16m4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tu( +// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tu( +// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tu( +// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tu( +// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tu( +// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tum( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tum( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tum( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tum( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tum( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tumu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tumu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tumu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tumu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vbfloat16mf4_t vs2, + __bf16 rs1, size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm, + vfloat32mf2_t vd, + vfloat32mf2_t vs2, + vbfloat16mf4_t vs1, + size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 1 x float> [[TMP0]] +// +vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd, + vfloat32mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vbfloat16mf2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, + vbfloat16mf2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 2 x float> [[TMP0]] +// +vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd, + vfloat32m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vbfloat16m1_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, + vbfloat16m1_t vs1, size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 4 x float> [[TMP0]] +// +vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd, + vfloat32m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vbfloat16m2_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, + vbfloat16m2_t vs1, size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 8 x float> [[TMP0]] +// +vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd, + vfloat32m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vbfloat16m4_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, + vbfloat16m4_t vs1, size_t vl) { + return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1) +// CHECK-RV64-NEXT: ret <vscale x 16 x float> [[TMP0]] +// +vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd, + vfloat32m8_t vs2, __bf16 rs1, + size_t vl) { + return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl); +} diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfeq.c new file mode 100644 index 0000000..3945f82 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfeq.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfeq_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfeq_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfeq_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfeq_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfeq_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfeq_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfge.c new file mode 100644 index 0000000..82586da --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfge.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfge_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfge_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfge_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfge_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfge_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfge_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfgt.c new file mode 100644 index 0000000..75ccbbc --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfgt.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfgt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfgt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfgt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfgt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfgt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfgt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfle.c new file mode 100644 index 0000000..49ff1c9 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfle.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfle_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfle_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfle_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfle_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfle_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfle_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmflt.c new file mode 100644 index 0000000..24b3f9c --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmflt.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmflt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmflt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmflt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmflt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmflt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmflt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfne.c new file mode 100644 index 0000000..ca3e134 --- /dev/null +++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfne.c @@ -0,0 +1,129 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// REQUIRES: riscv-registered-target +// RUN: %clang_cc1 -triple riscv64 -target-feature +v \ +// RUN: -target-feature +experimental-zvfbfa -disable-O0-optnone \ +// RUN: -emit-llvm %s -o - | opt -S -passes=mem2reg | \ +// RUN: FileCheck --check-prefix=CHECK-RV64 %s + +#include <riscv_vector.h> + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_mu( +// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 1 x i1> [[TMP0]] +// +vbool64_t test_vmfne_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_mu( +// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 2 x i1> [[TMP0]] +// +vbool32_t test_vmfne_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_mu( +// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 4 x i1> [[TMP0]] +// +vbool16_t test_vmfne_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_mu( +// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 8 x i1> [[TMP0]] +// +vbool8_t test_vmfne_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_mu( +// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 16 x i1> [[TMP0]] +// +vbool4_t test_vmfne_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + +// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_mu( +// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] { +// CHECK-RV64-NEXT: [[ENTRY:.*:]] +// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]]) +// CHECK-RV64-NEXT: ret <vscale x 32 x i1> [[TMP0]] +// +vbool2_t test_vmfne_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) { + return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl); +} + diff --git a/clang/test/CodeGen/X86/avx2-builtins.c b/clang/test/CodeGen/X86/avx2-builtins.c index b798618..a505d70 100644 --- a/clang/test/CodeGen/X86/avx2-builtins.c +++ b/clang/test/CodeGen/X86/avx2-builtins.c @@ -1038,6 +1038,7 @@ __m256i test_mm256_mulhrs_epi16(__m256i a, __m256i b) { // CHECK: call <16 x i16> @llvm.x86.avx2.pmul.hr.sw(<16 x i16> %{{.*}}, <16 x i16> %{{.*}}) return _mm256_mulhrs_epi16(a, b); } +TEST_CONSTEXPR(match_v16hi(_mm256_mulhrs_epi16((__m256i)(__v16hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600}, (__m256i)(__v16hi){+1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +5, -9, -13, +16, +18, -20, -21, -22, -22, +21, +20, -18, -16, +13, +9, -5)); __m256i test_mm256_mullo_epi16(__m256i a, __m256i b) { // CHECK-LABEL: test_mm256_mullo_epi16 diff --git a/clang/test/CodeGen/X86/avx512bw-builtins.c b/clang/test/CodeGen/X86/avx512bw-builtins.c index fddf17d..55bf482 100644 --- a/clang/test/CodeGen/X86/avx512bw-builtins.c +++ b/clang/test/CodeGen/X86/avx512bw-builtins.c @@ -1596,18 +1596,24 @@ __m512i test_mm512_mulhrs_epi16(__m512i __A, __m512i __B) { // CHECK: @llvm.x86.avx512.pmul.hr.sw.512 return _mm512_mulhrs_epi16(__A,__B); } +TEST_CONSTEXPR(match_v32hi(_mm512_mulhrs_epi16((__m512i)(__v32hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600, -1700, -1800, +1900, +2000, -2100, -2200, +2300, +2400, -2500, -2600, +2700, +2800, -2900, -3000, +3100, +3200}, (__m512i)(__v32hi){+3200, -3100, +3000, -2900, +2800, -2700, +2600, -2500, +2400, -2300, +2200, -2100, +2000, -1900, +1800, -1700, +1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +10, -19, -27, +35, +43, -49, -56, -61, -66, +70, +74, -77, -79, +81, +82, -83, -83, +82, +81, -79, -77, +74, +70, -66, -61, +56, +49, -43, -35, +27, +19, -10)); + __m512i test_mm512_mask_mulhrs_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B) { // CHECK-LABEL: test_mm512_mask_mulhrs_epi16 // CHECK: @llvm.x86.avx512.pmul.hr.sw.512 // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}} return _mm512_mask_mulhrs_epi16(__W,__U,__A,__B); } +TEST_CONSTEXPR(match_v32hi(_mm512_mask_mulhrs_epi16(_mm512_set1_epi16(1), 0x0000FFFF, (__m512i)(__v32hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600, -1700, -1800, +1900, +2000, -2100, -2200, +2300, +2400, -2500, -2600, +2700, +2800, -2900, -3000, +3100, +3200}, (__m512i)(__v32hi){+3200, -3100, +3000, -2900, +2800, -2700, +2600, -2500, +2400, -2300, +2200, -2100, +2000, -1900, +1800, -1700, +1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +10, -19, -27, +35, +43, -49, -56, -61, -66, +70, +74, -77, -79, +81, +82, -83, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1)); + __m512i test_mm512_maskz_mulhrs_epi16(__mmask32 __U, __m512i __A, __m512i __B) { // CHECK-LABEL: test_mm512_maskz_mulhrs_epi16 // CHECK: @llvm.x86.avx512.pmul.hr.sw.512 // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}} return _mm512_maskz_mulhrs_epi16(__U,__A,__B); } +TEST_CONSTEXPR(match_v32hi(_mm512_maskz_mulhrs_epi16(0x0000FFFF, (__m512i)(__v32hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600, -1700, -1800, +1900, +2000, -2100, -2200, +2300, +2400, -2500, -2600, +2700, +2800, -2900, -3000, +3100, +3200}, (__m512i)(__v32hi){+3200, -3100, +3000, -2900, +2800, -2700, +2600, -2500, +2400, -2300, +2200, -2100, +2000, -1900, +1800, -1700, +1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +10, -19, -27, +35, +43, -49, -56, -61, -66, +70, +74, -77, -79, +81, +82, -83, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)); + __m512i test_mm512_mulhi_epi16(__m512i __A, __m512i __B) { // CHECK-LABEL: test_mm512_mulhi_epi16 // CHECK: @llvm.x86.avx512.pmulh.w.512 diff --git a/clang/test/CodeGen/X86/avx512vlbw-builtins.c b/clang/test/CodeGen/X86/avx512vlbw-builtins.c index d569283..95e4d40 100644 --- a/clang/test/CodeGen/X86/avx512vlbw-builtins.c +++ b/clang/test/CodeGen/X86/avx512vlbw-builtins.c @@ -2061,6 +2061,7 @@ __m128i test_mm_mask_mulhrs_epi16(__m128i __W, __mmask8 __U, __m128i __X, __m128 // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}} return _mm_mask_mulhrs_epi16(__W, __U, __X, __Y); } +TEST_CONSTEXPR(match_v8hi(_mm_mask_mulhrs_epi16(_mm_set1_epi16(1), 0x0F, (__m128i)(__v8hi){+100, +200, -300, -400, +500, +600, -700, +800}, (__m128i)(__v8hi){+8000, -7000, +6000, -5000, +4000, -3000, +2000, -1000}), +24, -43, -55, +61, +1, +1, +1, +1)); __m128i test_mm_maskz_mulhrs_epi16(__mmask8 __U, __m128i __X, __m128i __Y) { // CHECK-LABEL: test_mm_maskz_mulhrs_epi16 @@ -2068,6 +2069,7 @@ __m128i test_mm_maskz_mulhrs_epi16(__mmask8 __U, __m128i __X, __m128i __Y) { // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}} return _mm_maskz_mulhrs_epi16(__U, __X, __Y); } +TEST_CONSTEXPR(match_v8hi(_mm_maskz_mulhrs_epi16(0x0F, (__m128i)(__v8hi){+100, +200, -300, -400, +500, +600, -700, +800}, (__m128i)(__v8hi){+8000, -7000, +6000, -5000, +4000, -3000, +2000, -1000}), +24, -43, -55, +61, 0, 0, 0, 0)); __m256i test_mm256_mask_mulhrs_epi16(__m256i __W, __mmask16 __U, __m256i __X, __m256i __Y) { // CHECK-LABEL: test_mm256_mask_mulhrs_epi16 @@ -2075,6 +2077,7 @@ __m256i test_mm256_mask_mulhrs_epi16(__m256i __W, __mmask16 __U, __m256i __X, __ // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}} return _mm256_mask_mulhrs_epi16(__W, __U, __X, __Y); } +TEST_CONSTEXPR(match_v16hi(_mm256_mask_mulhrs_epi16(_mm256_set1_epi16(1), 0xF00F, (__m256i)(__v16hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600}, (__m256i)(__v16hi){+1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +5, -9, -13, +16, +1, +1, +1, +1, +1, +1, +1, +1, -16, +13, +9, -5)); __m256i test_mm256_maskz_mulhrs_epi16(__mmask16 __U, __m256i __X, __m256i __Y) { // CHECK-LABEL: test_mm256_maskz_mulhrs_epi16 @@ -2082,6 +2085,7 @@ __m256i test_mm256_maskz_mulhrs_epi16(__mmask16 __U, __m256i __X, __m256i __Y) { // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}} return _mm256_maskz_mulhrs_epi16(__U, __X, __Y); } +TEST_CONSTEXPR(match_v16hi(_mm256_maskz_mulhrs_epi16(0xF00F, (__m256i)(__v16hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600}, (__m256i)(__v16hi){+1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +5, -9, -13, +16, 0, 0, 0, 0, 0, 0, 0, 0, -16, +13, +9, -5)); __m128i test_mm_mask_mulhi_epu16(__m128i __W, __mmask8 __U, __m128i __A, __m128i __B) { // CHECK-LABEL: test_mm_mask_mulhi_epu16 diff --git a/clang/test/CodeGen/X86/mmx-builtins.c b/clang/test/CodeGen/X86/mmx-builtins.c index d9041d4..c1ac57b 100644 --- a/clang/test/CodeGen/X86/mmx-builtins.c +++ b/clang/test/CodeGen/X86/mmx-builtins.c @@ -438,6 +438,7 @@ __m64 test_mm_mulhrs_pi16(__m64 a, __m64 b) { // CHECK: call <8 x i16> @llvm.x86.ssse3.pmul.hr.sw.128( return _mm_mulhrs_pi16(a, b); } +TEST_CONSTEXPR(match_v4hi(_mm_mulhrs_pi16((__m64)(__v4hi){+100, +200, -300, -400}, (__m64)(__v4hi){+30000, -20000, +10000, -5000}), +92, -122, -92, +61)); __m64 test_mm_mullo_pi16(__m64 a, __m64 b) { // CHECK-LABEL: test_mm_mullo_pi16 diff --git a/clang/test/CodeGen/X86/ssse3-builtins.c b/clang/test/CodeGen/X86/ssse3-builtins.c index 32abd9d..f70afc0 100644 --- a/clang/test/CodeGen/X86/ssse3-builtins.c +++ b/clang/test/CodeGen/X86/ssse3-builtins.c @@ -110,6 +110,7 @@ __m128i test_mm_mulhrs_epi16(__m128i a, __m128i b) { // CHECK: call <8 x i16> @llvm.x86.ssse3.pmul.hr.sw.128(<8 x i16> %{{.*}}, <8 x i16> %{{.*}}) return _mm_mulhrs_epi16(a, b); } +TEST_CONSTEXPR(match_v8hi(_mm_mulhrs_epi16((__m128i)(__v8hi){+100, +200, -300, -400, +500, +600, -700, +800}, (__m128i)(__v8hi){+8000, -7000, +6000, -5000, +4000, -3000, +2000, -1000}), +24, -43, -55, +61, +61, -55, -43, -24)); __m128i test_mm_shuffle_epi8(__m128i a, __m128i b) { // CHECK-LABEL: test_mm_shuffle_epi8 |