diff options
Diffstat (limited to 'llvm/test')
90 files changed, 3421 insertions, 2627 deletions
diff --git a/llvm/test/Analysis/LoopAccessAnalysis/early-exit-runtime-checks.ll b/llvm/test/Analysis/LoopAccessAnalysis/early-exit-runtime-checks.ll index 6d9aa8d..05aad8a 100644 --- a/llvm/test/Analysis/LoopAccessAnalysis/early-exit-runtime-checks.ll +++ b/llvm/test/Analysis/LoopAccessAnalysis/early-exit-runtime-checks.ll @@ -770,10 +770,10 @@ define void @all_exits_dominate_latch_countable_exits_at_most_500_iterations_kno ; CHECK-NEXT: %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv ; CHECK-NEXT: Grouped accesses: ; CHECK-NEXT: Group GRP0: -; CHECK-NEXT: (Low: %B High: inttoptr (i64 -1 to ptr)) +; CHECK-NEXT: (Low: %B High: (2000 + %B)) ; CHECK-NEXT: Member: {%B,+,4}<nuw><%loop.header> ; CHECK-NEXT: Group GRP1: -; CHECK-NEXT: (Low: %A High: inttoptr (i64 -1 to ptr)) +; CHECK-NEXT: (Low: %A High: (2000 + %A)) ; CHECK-NEXT: Member: {%A,+,4}<nuw><%loop.header> ; CHECK-EMPTY: ; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop. diff --git a/llvm/test/Bitcode/DW_OP_rot_neg_abs.ll b/llvm/test/Bitcode/DW_OP_rot_neg_abs.ll new file mode 100644 index 0000000..e185530 --- /dev/null +++ b/llvm/test/Bitcode/DW_OP_rot_neg_abs.ll @@ -0,0 +1,10 @@ +;; This test checks the validity of DWARF operators DW_OP_rot, DW_OP_neg, and DW_OP_abs. + +; RUN: llvm-as < %s | llvm-dis | llvm-as | llvm-dis | FileCheck %s + +; CHECK: !DIExpression(DW_OP_push_object_address, DW_OP_lit0, DW_OP_lit0, DW_OP_neg, DW_OP_abs, DW_OP_rot, DW_OP_rot, DW_OP_rot, DW_OP_plus, DW_OP_plus) + +; ModuleID = 'DW_OP_rot_neg_abs.adb' +source_filename = "/dir/DW_OP_rot_neg_abs.ll" + +!named = !{!DIExpression(DW_OP_push_object_address, DW_OP_lit0, DW_OP_lit0, DW_OP_neg, DW_OP_abs, DW_OP_rot, DW_OP_rot, DW_OP_rot, DW_OP_plus, DW_OP_plus)} diff --git a/llvm/test/CodeGen/AArch64/sve-int-mul-neg.ll b/llvm/test/CodeGen/AArch64/sve-int-mul-neg.ll new file mode 100644 index 0000000..a1065bc --- /dev/null +++ b/llvm/test/CodeGen/AArch64/sve-int-mul-neg.ll @@ -0,0 +1,131 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5 +; RUN: llc -verify-machineinstrs -mattr=+sve < %s | FileCheck %s + +target triple = "aarch64-unknown-linux-gnu" + +; Muls with (-1) as operand should fold to neg. +define <vscale x 16 x i8> @mul_neg_fold_i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) { +; CHECK-LABEL: mul_neg_fold_i8: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.b, p0/m, z0.b +; CHECK-NEXT: ret + %1 = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a, <vscale x 16 x i8> splat(i8 -1)) + ret <vscale x 16 x i8> %1 +} + +define <vscale x 8 x i16> @mul_neg_fold_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) { +; CHECK-LABEL: mul_neg_fold_i16: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.h, p0/m, z0.h +; CHECK-NEXT: ret + %1 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a, <vscale x 8 x i16> splat(i16 -1)) + ret <vscale x 8 x i16> %1 +} + +define <vscale x 4 x i32> @mul_neg_fold_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) { +; CHECK-LABEL: mul_neg_fold_i32: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.s, p0/m, z0.s +; CHECK-NEXT: ret + %1 = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a, <vscale x 4 x i32> splat(i32 -1)) + ret <vscale x 4 x i32> %1 +} + +define <vscale x 2 x i64> @mul_neg_fold_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) { +; CHECK-LABEL: mul_neg_fold_i64: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.d, p0/m, z0.d +; CHECK-NEXT: ret + %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a, <vscale x 2 x i64> splat(i64 -1)) + ret <vscale x 2 x i64> %1 +} + +define <vscale x 16 x i8> @mul_neg_fold_u_i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) { +; CHECK-LABEL: mul_neg_fold_u_i8: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.b, p0/m, z0.b +; CHECK-NEXT: ret + %1 = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a, <vscale x 16 x i8> splat(i8 -1)) + ret <vscale x 16 x i8> %1 +} + +define <vscale x 8 x i16> @mul_neg_fold_u_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) { +; CHECK-LABEL: mul_neg_fold_u_i16: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.h, p0/m, z0.h +; CHECK-NEXT: ret + %1 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.u.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a, <vscale x 8 x i16> splat(i16 -1)) + ret <vscale x 8 x i16> %1 +} + +define <vscale x 4 x i32> @mul_neg_fold_u_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) { +; CHECK-LABEL: mul_neg_fold_u_i32: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.s, p0/m, z0.s +; CHECK-NEXT: ret + %1 = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.u.nxv4i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a, <vscale x 4 x i32> splat(i32 -1)) + ret <vscale x 4 x i32> %1 +} + +define <vscale x 2 x i64> @mul_neg_fold_u_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) { +; CHECK-LABEL: mul_neg_fold_u_i64: +; CHECK: // %bb.0: +; CHECK-NEXT: neg z0.d, p0/m, z0.d +; CHECK-NEXT: ret + %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.u.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a, <vscale x 2 x i64> splat(i64 -1)) + ret <vscale x 2 x i64> %1 +} + +define <vscale x 16 x i8> @mul_neg_fold_different_argument_order_i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) { +; CHECK-LABEL: mul_neg_fold_different_argument_order_i8: +; CHECK: // %bb.0: +; CHECK-NEXT: mov z1.b, #-1 // =0xffffffffffffffff +; CHECK-NEXT: neg z1.b, p0/m, z0.b +; CHECK-NEXT: mov z0.d, z1.d +; CHECK-NEXT: ret + %1 = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> splat(i8 -1), <vscale x 16 x i8> %a) + ret <vscale x 16 x i8> %1 +} + +define <vscale x 8 x i16> @mul_neg_fold_different_argument_order_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) { +; CHECK-LABEL: mul_neg_fold_different_argument_order_i16: +; CHECK: // %bb.0: +; CHECK-NEXT: mov z1.h, #-1 // =0xffffffffffffffff +; CHECK-NEXT: neg z1.h, p0/m, z0.h +; CHECK-NEXT: mov z0.d, z1.d +; CHECK-NEXT: ret + %1 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> splat(i16 -1), <vscale x 8 x i16> %a) + ret <vscale x 8 x i16> %1 +} + +define <vscale x 4 x i32> @mul_neg_fold_different_argument_order_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) { +; CHECK-LABEL: mul_neg_fold_different_argument_order_i32: +; CHECK: // %bb.0: +; CHECK-NEXT: mov z1.s, #-1 // =0xffffffffffffffff +; CHECK-NEXT: neg z1.s, p0/m, z0.s +; CHECK-NEXT: mov z0.d, z1.d +; CHECK-NEXT: ret + %1 = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> splat(i32 -1), <vscale x 4 x i32> %a) + ret <vscale x 4 x i32> %1 +} + +define <vscale x 2 x i64> @mul_neg_fold_different_argument_order_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) { +; CHECK-LABEL: mul_neg_fold_different_argument_order_i64: +; CHECK: // %bb.0: +; CHECK-NEXT: mov z1.d, #-1 // =0xffffffffffffffff +; CHECK-NEXT: neg z1.d, p0/m, z0.d +; CHECK-NEXT: mov z0.d, z1.d +; CHECK-NEXT: ret + %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> splat(i64 -1), <vscale x 2 x i64> %a) + ret <vscale x 2 x i64> %1 +} + +declare <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>) +declare <vscale x 8 x i16> @llvm.aarch64.sve.mul.u.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>) +declare <vscale x 4 x i32> @llvm.aarch64.sve.mul.u.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>) +declare <vscale x 2 x i64> @llvm.aarch64.sve.mul.u.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>) + +declare <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>) +declare <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>) +declare <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>) +declare <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>) diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/load-constant.96.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/load-constant.96.ll index 41fda6d..efa51ea 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/load-constant.96.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/load-constant.96.ll @@ -90,26 +90,24 @@ define <3 x i32> @v_load_constant_v3i32_align1(ptr addrspace(4) %ptr) { ; GFX1250-NOUNALIGNED-NEXT: global_load_u8 v10, v[0:1], off offset:8 ; GFX1250-NOUNALIGNED-NEXT: global_load_u8 v11, v[0:1], off offset:9 ; GFX1250-NOUNALIGNED-NEXT: global_load_u8 v12, v[0:1], off offset:11 -; GFX1250-NOUNALIGNED-NEXT: global_load_u8 v0, v[0:1], off offset:10 +; GFX1250-NOUNALIGNED-NEXT: global_load_u8 v13, v[0:1], off offset:10 ; GFX1250-NOUNALIGNED-NEXT: s_wait_loadcnt 0xa ; GFX1250-NOUNALIGNED-NEXT: s_wait_xcnt 0x0 -; GFX1250-NOUNALIGNED-NEXT: v_lshl_or_b32 v1, v3, 8, v2 +; GFX1250-NOUNALIGNED-NEXT: v_lshl_or_b32 v0, v3, 8, v2 ; GFX1250-NOUNALIGNED-NEXT: s_wait_loadcnt 0x8 -; GFX1250-NOUNALIGNED-NEXT: v_dual_lshlrev_b32 v3, 16, v4 :: v_dual_lshlrev_b32 v2, 24, v5 +; GFX1250-NOUNALIGNED-NEXT: v_dual_lshlrev_b32 v2, 16, v4 :: v_dual_lshlrev_b32 v1, 24, v5 ; GFX1250-NOUNALIGNED-NEXT: s_wait_loadcnt 0x6 -; GFX1250-NOUNALIGNED-NEXT: v_lshl_or_b32 v4, v7, 8, v6 +; GFX1250-NOUNALIGNED-NEXT: v_lshl_or_b32 v3, v7, 8, v6 ; GFX1250-NOUNALIGNED-NEXT: s_wait_loadcnt 0x4 -; GFX1250-NOUNALIGNED-NEXT: v_dual_lshlrev_b32 v6, 16, v8 :: v_dual_lshlrev_b32 v5, 24, v9 +; GFX1250-NOUNALIGNED-NEXT: v_dual_lshlrev_b32 v5, 16, v8 :: v_dual_lshlrev_b32 v4, 24, v9 +; GFX1250-NOUNALIGNED-NEXT: v_or3_b32 v0, v1, v2, v0 ; GFX1250-NOUNALIGNED-NEXT: s_wait_loadcnt 0x2 -; GFX1250-NOUNALIGNED-NEXT: v_lshl_or_b32 v7, v11, 8, v10 -; GFX1250-NOUNALIGNED-NEXT: s_wait_loadcnt 0x1 -; GFX1250-NOUNALIGNED-NEXT: v_lshlrev_b32_e32 v8, 24, v12 +; GFX1250-NOUNALIGNED-NEXT: v_lshl_or_b32 v6, v11, 8, v10 ; GFX1250-NOUNALIGNED-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NOUNALIGNED-NEXT: v_lshlrev_b32_e32 v9, 16, v0 -; GFX1250-NOUNALIGNED-NEXT: v_or3_b32 v0, v2, v3, v1 -; GFX1250-NOUNALIGNED-NEXT: v_or3_b32 v1, v5, v6, v4 -; GFX1250-NOUNALIGNED-NEXT: s_delay_alu instid0(VALU_DEP_3) -; GFX1250-NOUNALIGNED-NEXT: v_or3_b32 v2, v8, v9, v7 +; GFX1250-NOUNALIGNED-NEXT: v_dual_lshlrev_b32 v7, 24, v12 :: v_dual_lshlrev_b32 v8, 16, v13 +; GFX1250-NOUNALIGNED-NEXT: v_or3_b32 v1, v4, v5, v3 +; GFX1250-NOUNALIGNED-NEXT: s_delay_alu instid0(VALU_DEP_2) +; GFX1250-NOUNALIGNED-NEXT: v_or3_b32 v2, v7, v8, v6 ; GFX1250-NOUNALIGNED-NEXT: s_set_pc_i64 s[30:31] ; ; GFX9-UNALIGNED-LABEL: v_load_constant_v3i32_align1: @@ -942,7 +940,7 @@ define amdgpu_ps <3 x i32> @s_load_constant_v3i32_align1(ptr addrspace(4) inreg ; ; GFX1250-NOUNALIGNED-LABEL: s_load_constant_v3i32_align1: ; GFX1250-NOUNALIGNED: ; %bb.0: -; GFX1250-NOUNALIGNED-NEXT: s_clause 0xa +; GFX1250-NOUNALIGNED-NEXT: s_clause 0xb ; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s2, s[0:1], 0x1 ; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s3, s[0:1], 0x3 ; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s4, s[0:1], 0x2 @@ -954,27 +952,26 @@ define amdgpu_ps <3 x i32> @s_load_constant_v3i32_align1(ptr addrspace(4) inreg ; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s10, s[0:1], 0x0 ; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s11, s[0:1], 0x4 ; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s12, s[0:1], 0xa -; GFX1250-NOUNALIGNED-NEXT: s_wait_xcnt 0x0 -; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s1, s[0:1], 0x8 +; GFX1250-NOUNALIGNED-NEXT: s_load_u8 s13, s[0:1], 0x8 ; GFX1250-NOUNALIGNED-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s0, s2, 8 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s2, s3, 24 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s3, s4, 16 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s4, s5, 8 -; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s2, s2, s3 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s5, s6, 24 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s6, s7, 16 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s7, s8, 8 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s1, s3, 24 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s2, s4, 16 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s3, s5, 8 +; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s1, s1, s2 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s4, s6, 24 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s5, s7, 16 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s6, s8, 8 ; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s0, s0, s10 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s8, s9, 24 -; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s0, s2, s0 -; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s2, s12, 16 -; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s3, s4, s11 -; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s4, s5, s6 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s7, s9, 24 +; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s0, s1, s0 +; GFX1250-NOUNALIGNED-NEXT: s_lshl_b32 s1, s12, 16 +; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s2, s3, s11 +; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s3, s4, s5 +; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s4, s6, s13 ; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s5, s7, s1 -; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s2, s8, s2 -; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s1, s4, s3 -; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s2, s2, s5 +; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s1, s3, s2 +; GFX1250-NOUNALIGNED-NEXT: s_or_b32 s2, s5, s4 ; GFX1250-NOUNALIGNED-NEXT: ; return to shader part epilog ; ; GFX9-UNALIGNED-LABEL: s_load_constant_v3i32_align1: @@ -1351,11 +1348,25 @@ define amdgpu_ps <3 x i32> @s_load_constant_v3i32_align2(ptr addrspace(4) inreg } define amdgpu_ps <3 x i32> @s_load_constant_v3i32_align4(ptr addrspace(4) inreg %ptr) { -; GFX12-LABEL: s_load_constant_v3i32_align4: -; GFX12: ; %bb.0: -; GFX12-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 -; GFX12-NEXT: s_wait_kmcnt 0x0 -; GFX12-NEXT: ; return to shader part epilog +; GFX12-UNALIGNED-LABEL: s_load_constant_v3i32_align4: +; GFX12-UNALIGNED: ; %bb.0: +; GFX12-UNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-UNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-UNALIGNED-NEXT: ; return to shader part epilog +; +; GFX12-NOUNALIGNED-LABEL: s_load_constant_v3i32_align4: +; GFX12-NOUNALIGNED: ; %bb.0: +; GFX12-NOUNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-NOUNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-NOUNALIGNED-NEXT: ; return to shader part epilog +; +; GFX1250-LABEL: s_load_constant_v3i32_align4: +; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_mov_b32 s4, s0 +; GFX1250-NEXT: s_mov_b32 s5, s1 +; GFX1250-NEXT: s_load_b96 s[0:2], s[4:5], 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: ; return to shader part epilog ; ; GFX9-LABEL: s_load_constant_v3i32_align4: ; GFX9: ; %bb.0: @@ -1388,11 +1399,25 @@ define amdgpu_ps <3 x i32> @s_load_constant_v3i32_align4(ptr addrspace(4) inreg } define amdgpu_ps i96 @s_load_constant_i96_align8(ptr addrspace(4) inreg %ptr) { -; GFX12-LABEL: s_load_constant_i96_align8: -; GFX12: ; %bb.0: -; GFX12-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 -; GFX12-NEXT: s_wait_kmcnt 0x0 -; GFX12-NEXT: ; return to shader part epilog +; GFX12-UNALIGNED-LABEL: s_load_constant_i96_align8: +; GFX12-UNALIGNED: ; %bb.0: +; GFX12-UNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-UNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-UNALIGNED-NEXT: ; return to shader part epilog +; +; GFX12-NOUNALIGNED-LABEL: s_load_constant_i96_align8: +; GFX12-NOUNALIGNED: ; %bb.0: +; GFX12-NOUNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-NOUNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-NOUNALIGNED-NEXT: ; return to shader part epilog +; +; GFX1250-LABEL: s_load_constant_i96_align8: +; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_mov_b32 s4, s0 +; GFX1250-NEXT: s_mov_b32 s5, s1 +; GFX1250-NEXT: s_load_b96 s[0:2], s[4:5], 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: ; return to shader part epilog ; ; GFX9-LABEL: s_load_constant_i96_align8: ; GFX9: ; %bb.0: @@ -1425,11 +1450,25 @@ define amdgpu_ps i96 @s_load_constant_i96_align8(ptr addrspace(4) inreg %ptr) { } define amdgpu_ps <3 x i32> @s_load_constant_v3i32_align8(ptr addrspace(4) inreg %ptr) { -; GFX12-LABEL: s_load_constant_v3i32_align8: -; GFX12: ; %bb.0: -; GFX12-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 -; GFX12-NEXT: s_wait_kmcnt 0x0 -; GFX12-NEXT: ; return to shader part epilog +; GFX12-UNALIGNED-LABEL: s_load_constant_v3i32_align8: +; GFX12-UNALIGNED: ; %bb.0: +; GFX12-UNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-UNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-UNALIGNED-NEXT: ; return to shader part epilog +; +; GFX12-NOUNALIGNED-LABEL: s_load_constant_v3i32_align8: +; GFX12-NOUNALIGNED: ; %bb.0: +; GFX12-NOUNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-NOUNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-NOUNALIGNED-NEXT: ; return to shader part epilog +; +; GFX1250-LABEL: s_load_constant_v3i32_align8: +; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_mov_b32 s4, s0 +; GFX1250-NEXT: s_mov_b32 s5, s1 +; GFX1250-NEXT: s_load_b96 s[0:2], s[4:5], 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: ; return to shader part epilog ; ; GFX9-LABEL: s_load_constant_v3i32_align8: ; GFX9: ; %bb.0: @@ -1462,11 +1501,25 @@ define amdgpu_ps <3 x i32> @s_load_constant_v3i32_align8(ptr addrspace(4) inreg } define amdgpu_ps <3 x i32> @s_load_constant_v6i16_align8(ptr addrspace(4) inreg %ptr) { -; GFX12-LABEL: s_load_constant_v6i16_align8: -; GFX12: ; %bb.0: -; GFX12-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 -; GFX12-NEXT: s_wait_kmcnt 0x0 -; GFX12-NEXT: ; return to shader part epilog +; GFX12-UNALIGNED-LABEL: s_load_constant_v6i16_align8: +; GFX12-UNALIGNED: ; %bb.0: +; GFX12-UNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-UNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-UNALIGNED-NEXT: ; return to shader part epilog +; +; GFX12-NOUNALIGNED-LABEL: s_load_constant_v6i16_align8: +; GFX12-NOUNALIGNED: ; %bb.0: +; GFX12-NOUNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-NOUNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-NOUNALIGNED-NEXT: ; return to shader part epilog +; +; GFX1250-LABEL: s_load_constant_v6i16_align8: +; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_mov_b32 s4, s0 +; GFX1250-NEXT: s_mov_b32 s5, s1 +; GFX1250-NEXT: s_load_b96 s[0:2], s[4:5], 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: ; return to shader part epilog ; ; GFX9-LABEL: s_load_constant_v6i16_align8: ; GFX9: ; %bb.0: @@ -1500,24 +1553,64 @@ define amdgpu_ps <3 x i32> @s_load_constant_v6i16_align8(ptr addrspace(4) inreg } define amdgpu_ps <12 x i8> @s_load_constant_v12i8_align8(ptr addrspace(4) inreg %ptr) { -; GFX12-LABEL: s_load_constant_v12i8_align8: -; GFX12: ; %bb.0: -; GFX12-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 -; GFX12-NEXT: s_wait_kmcnt 0x0 -; GFX12-NEXT: s_lshr_b32 s13, s0, 8 -; GFX12-NEXT: s_lshr_b32 s12, s0, 16 -; GFX12-NEXT: s_lshr_b32 s3, s0, 24 -; GFX12-NEXT: s_lshr_b32 s5, s1, 8 -; GFX12-NEXT: s_lshr_b32 s6, s1, 16 -; GFX12-NEXT: s_lshr_b32 s7, s1, 24 -; GFX12-NEXT: s_lshr_b32 s9, s2, 8 -; GFX12-NEXT: s_lshr_b32 s10, s2, 16 -; GFX12-NEXT: s_lshr_b32 s11, s2, 24 -; GFX12-NEXT: s_mov_b32 s4, s1 -; GFX12-NEXT: s_mov_b32 s8, s2 -; GFX12-NEXT: s_mov_b32 s1, s13 -; GFX12-NEXT: s_mov_b32 s2, s12 -; GFX12-NEXT: ; return to shader part epilog +; GFX12-UNALIGNED-LABEL: s_load_constant_v12i8_align8: +; GFX12-UNALIGNED: ; %bb.0: +; GFX12-UNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-UNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s13, s0, 8 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s12, s0, 16 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s3, s0, 24 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s5, s1, 8 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s6, s1, 16 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s7, s1, 24 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s9, s2, 8 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s10, s2, 16 +; GFX12-UNALIGNED-NEXT: s_lshr_b32 s11, s2, 24 +; GFX12-UNALIGNED-NEXT: s_mov_b32 s4, s1 +; GFX12-UNALIGNED-NEXT: s_mov_b32 s8, s2 +; GFX12-UNALIGNED-NEXT: s_mov_b32 s1, s13 +; GFX12-UNALIGNED-NEXT: s_mov_b32 s2, s12 +; GFX12-UNALIGNED-NEXT: ; return to shader part epilog +; +; GFX12-NOUNALIGNED-LABEL: s_load_constant_v12i8_align8: +; GFX12-NOUNALIGNED: ; %bb.0: +; GFX12-NOUNALIGNED-NEXT: s_load_b96 s[0:2], s[0:1], 0x0 +; GFX12-NOUNALIGNED-NEXT: s_wait_kmcnt 0x0 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s13, s0, 8 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s12, s0, 16 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s3, s0, 24 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s5, s1, 8 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s6, s1, 16 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s7, s1, 24 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s9, s2, 8 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s10, s2, 16 +; GFX12-NOUNALIGNED-NEXT: s_lshr_b32 s11, s2, 24 +; GFX12-NOUNALIGNED-NEXT: s_mov_b32 s4, s1 +; GFX12-NOUNALIGNED-NEXT: s_mov_b32 s8, s2 +; GFX12-NOUNALIGNED-NEXT: s_mov_b32 s1, s13 +; GFX12-NOUNALIGNED-NEXT: s_mov_b32 s2, s12 +; GFX12-NOUNALIGNED-NEXT: ; return to shader part epilog +; +; GFX1250-LABEL: s_load_constant_v12i8_align8: +; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_mov_b32 s4, s0 +; GFX1250-NEXT: s_mov_b32 s5, s1 +; GFX1250-NEXT: s_load_b96 s[0:2], s[4:5], 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: s_lshr_b32 s13, s0, 8 +; GFX1250-NEXT: s_lshr_b32 s12, s0, 16 +; GFX1250-NEXT: s_lshr_b32 s3, s0, 24 +; GFX1250-NEXT: s_lshr_b32 s5, s1, 8 +; GFX1250-NEXT: s_lshr_b32 s6, s1, 16 +; GFX1250-NEXT: s_lshr_b32 s7, s1, 24 +; GFX1250-NEXT: s_lshr_b32 s9, s2, 8 +; GFX1250-NEXT: s_lshr_b32 s10, s2, 16 +; GFX1250-NEXT: s_lshr_b32 s11, s2, 24 +; GFX1250-NEXT: s_mov_b32 s4, s1 +; GFX1250-NEXT: s_mov_b32 s8, s2 +; GFX1250-NEXT: s_mov_b32 s1, s13 +; GFX1250-NEXT: s_mov_b32 s2, s12 +; GFX1250-NEXT: ; return to shader part epilog ; ; GFX9-LABEL: s_load_constant_v12i8_align8: ; GFX9: ; %bb.0: diff --git a/llvm/test/CodeGen/AMDGPU/addrspacecast-gas.ll b/llvm/test/CodeGen/AMDGPU/addrspacecast-gas.ll index aac499f..b486fabb 100644 --- a/llvm/test/CodeGen/AMDGPU/addrspacecast-gas.ll +++ b/llvm/test/CodeGen/AMDGPU/addrspacecast-gas.ll @@ -9,15 +9,14 @@ target triple = "amdgcn-amd-amdhsa" define amdgpu_kernel void @use_private_to_flat_addrspacecast(ptr addrspace(5) %ptr) { ; GFX1250-SDAG-LABEL: use_private_to_flat_addrspacecast: ; GFX1250-SDAG: ; %bb.0: -; GFX1250-SDAG-NEXT: s_load_b32 s2, s[4:5], 0x24 +; GFX1250-SDAG-NEXT: s_load_b32 s0, s[4:5], 0x24 ; GFX1250-SDAG-NEXT: v_mbcnt_lo_u32_b32 v0, -1, 0 -; GFX1250-SDAG-NEXT: s_mov_b64 s[0:1], src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_lshlrev_b32 v1, 20, v0 -; GFX1250-SDAG-NEXT: s_cmp_lg_u32 s2, -1 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_lshlrev_b32 v1, 20, v0 +; GFX1250-SDAG-NEXT: s_cmp_lg_u32 s0, -1 ; GFX1250-SDAG-NEXT: s_cselect_b32 vcc_lo, -1, 0 -; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] +; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], src_flat_scratch_base_lo, v[0:1] ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_cndmask_b32 v1, 0, v1 ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v0, 0, v0, vcc_lo @@ -56,13 +55,11 @@ define amdgpu_kernel void @use_private_to_flat_addrspacecast_nonnull(ptr addrspa ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: s_load_b32 s0, s[4:5], 0x24 ; GFX1250-SDAG-NEXT: v_mbcnt_lo_u32_b32 v0, -1, 0 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_lshlrev_b32 v1, 20, v0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v0, s0 -; GFX1250-SDAG-NEXT: s_mov_b64 s[0:1], src_flat_scratch_base_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] +; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], src_flat_scratch_base_lo, v[0:1] ; GFX1250-SDAG-NEXT: flat_store_b32 v[0:1], v2 scope:SCOPE_SYS ; GFX1250-SDAG-NEXT: s_wait_storecnt 0x0 ; GFX1250-SDAG-NEXT: s_endpgm @@ -91,10 +88,9 @@ define amdgpu_kernel void @use_flat_to_private_addrspacecast(ptr %ptr) { ; GFX1250-LABEL: use_flat_to_private_addrspacecast: ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 -; GFX1250-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo ; GFX1250-NEXT: v_mov_b32_e32 v0, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_sub_co_i32 s2, s0, s2 +; GFX1250-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo ; GFX1250-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-NEXT: s_cselect_b32 s0, s2, -1 ; GFX1250-NEXT: scratch_store_b32 off, v0, s0 scope:SCOPE_SYS @@ -110,9 +106,8 @@ define amdgpu_kernel void @use_flat_to_private_addrspacecast_nonnull(ptr %ptr) { ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: s_load_b32 s0, s[4:5], 0x24 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v0, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 -; GFX1250-SDAG-NEXT: s_sub_co_i32 s0, s0, s1 +; GFX1250-SDAG-NEXT: s_sub_co_i32 s0, s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: scratch_store_b32 off, v0, s0 scope:SCOPE_SYS ; GFX1250-SDAG-NEXT: s_wait_storecnt 0x0 ; GFX1250-SDAG-NEXT: s_endpgm @@ -122,9 +117,7 @@ define amdgpu_kernel void @use_flat_to_private_addrspacecast_nonnull(ptr %ptr) { ; GFX1250-GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 ; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, 0 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 -; GFX1250-GISEL-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo -; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s0, s1 +; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: scratch_store_b32 off, v0, s0 scope:SCOPE_SYS ; GFX1250-GISEL-NEXT: s_wait_storecnt 0x0 ; GFX1250-GISEL-NEXT: s_endpgm diff --git a/llvm/test/CodeGen/AMDGPU/amdgpu-attributor-flat-scratch-init-asan.ll b/llvm/test/CodeGen/AMDGPU/amdgpu-attributor-flat-scratch-init-asan.ll new file mode 100644 index 0000000..0d68762 --- /dev/null +++ b/llvm/test/CodeGen/AMDGPU/amdgpu-attributor-flat-scratch-init-asan.ll @@ -0,0 +1,24 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-attributes --check-globals all --version 6 +; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -passes='amdgpu-attributor' %s -o - | FileCheck %s + +@lds_1 = internal addrspace(3) global [1 x i8] poison, align 4 + +;. +; CHECK: @lds_1 = internal addrspace(3) global [1 x i8] poison, align 4 +;. +define amdgpu_kernel void @k0() #0 { +; CHECK: Function Attrs: sanitize_address +; CHECK-LABEL: define amdgpu_kernel void @k0( +; CHECK-SAME: ) #[[ATTR0:[0-9]+]] { +; CHECK-NEXT: store i8 7, ptr addrspace(3) @lds_1, align 4 +; CHECK-NEXT: ret void +; + store i8 7, ptr addrspace(3) @lds_1, align 4 + ret void +} + +attributes #0 = { sanitize_address } +; "amdgpu-no-flat-scratch-init" attribute should not be present in attribute list +;. +; CHECK: attributes #[[ATTR0]] = { sanitize_address "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "target-cpu"="gfx90a" "uniform-work-group-size"="false" } +;. diff --git a/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll b/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll index a688b6f..fb566e5 100644 --- a/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll +++ b/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll @@ -707,8 +707,8 @@ attributes #6 = { "enqueued-block" } ; ATTRIBUTOR_HSA: attributes #[[ATTR14]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "target-cpu"="gfx900" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR15:[0-9]+]] = { nounwind "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR16]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } -; ATTRIBUTOR_HSA: attributes #[[ATTR17]] = { nounwind sanitize_address "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } -; ATTRIBUTOR_HSA: attributes #[[ATTR18]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } +; ATTRIBUTOR_HSA: attributes #[[ATTR17]] = { nounwind sanitize_address "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } +; ATTRIBUTOR_HSA: attributes #[[ATTR18]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR19:[0-9]+]] = { nounwind sanitize_address "amdgpu-no-implicitarg-ptr" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR20:[0-9]+]] = { "enqueued-block" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR21]] = { "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "enqueued-block" "uniform-work-group-size"="false" } diff --git a/llvm/test/CodeGen/AMDGPU/atomics-system-scope.ll b/llvm/test/CodeGen/AMDGPU/atomics-system-scope.ll index ef52694..54871a6 100644 --- a/llvm/test/CodeGen/AMDGPU/atomics-system-scope.ll +++ b/llvm/test/CodeGen/AMDGPU/atomics-system-scope.ll @@ -538,58 +538,61 @@ define double @flat_system_atomic_fadd_f64(ptr %ptr, double %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: v_dual_mov_b32 v5, v1 :: v_dual_mov_b32 v4, v0 ; GFX1250-NEXT: s_mov_b64 s[0:1], src_shared_base ; GFX1250-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 -; GFX1250-NEXT: v_cmpx_ne_u32_e64 s1, v1 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-NEXT: v_cmpx_ne_u32_e64 s1, v5 ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB34_6 -; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.check.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s1, v1 -; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 +; GFX1250-NEXT: s_cbranch_execnz .LBB34_3 +; GFX1250-NEXT: ; %bb.1: ; %Flow2 +; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 +; GFX1250-NEXT: s_cbranch_execnz .LBB34_8 +; GFX1250-NEXT: .LBB34_2: ; %atomicrmw.phi +; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 +; GFX1250-NEXT: s_wait_loadcnt 0x0 +; GFX1250-NEXT: s_set_pc_i64 s[30:31] +; GFX1250-NEXT: .LBB34_3: ; %atomicrmw.check.private +; GFX1250-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-NEXT: s_and_saveexec_b32 s1, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s1, exec_lo, s1 -; GFX1250-NEXT: s_cbranch_execz .LBB34_3 -; GFX1250-NEXT: ; %bb.2: ; %atomicrmw.global -; GFX1250-NEXT: global_atomic_add_f64 v[4:5], v[0:1], v[2:3], off th:TH_ATOMIC_RETURN scope:SCOPE_SYS -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_cbranch_execz .LBB34_5 +; GFX1250-NEXT: ; %bb.4: ; %atomicrmw.global +; GFX1250-NEXT: global_atomic_add_f64 v[0:1], v[4:5], v[2:3], off th:TH_ATOMIC_RETURN scope:SCOPE_SYS +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB34_3: ; %Flow +; GFX1250-NEXT: .LBB34_5: ; %Flow ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s1, s1 -; GFX1250-NEXT: s_cbranch_execz .LBB34_5 -; GFX1250-NEXT: ; %bb.4: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] +; GFX1250-NEXT: s_cbranch_execz .LBB34_7 +; GFX1250-NEXT: ; %bb.6: ; %atomicrmw.private +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s2, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo -; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off +; GFX1250-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo +; GFX1250-NEXT: scratch_load_b64 v[0:1], v4, off ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_add_f64_e32 v[0:1], v[4:5], v[2:3] -; GFX1250-NEXT: scratch_store_b64 v6, v[0:1], off -; GFX1250-NEXT: .LBB34_5: ; %Flow1 +; GFX1250-NEXT: v_add_f64_e32 v[2:3], v[0:1], v[2:3] +; GFX1250-NEXT: scratch_store_b64 v4, v[2:3], off +; GFX1250-NEXT: .LBB34_7: ; %Flow1 ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s1 -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB34_6: ; %Flow2 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB34_8 -; GFX1250-NEXT: ; %bb.7: ; %atomicrmw.shared -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-NEXT: v_cndmask_b32_e32 v0, -1, v0, vcc_lo +; GFX1250-NEXT: s_cbranch_execz .LBB34_2 +; GFX1250-NEXT: .LBB34_8: ; %atomicrmw.shared +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: ds_add_rtn_f64 v[4:5], v0, v[2:3] -; GFX1250-NEXT: .LBB34_8: ; %atomicrmw.phi +; GFX1250-NEXT: v_cndmask_b32_e32 v0, -1, v4, vcc_lo +; GFX1250-NEXT: ds_add_rtn_f64 v[0:1], v0, v[2:3] ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 -; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, v4 :: v_dual_mov_b32 v1, v5 +; GFX1250-NEXT: s_wait_dscnt 0x0 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %result = atomicrmw fadd ptr %ptr, double %val monotonic ret double %result @@ -600,58 +603,61 @@ define double @flat_one_as_atomic_fadd_f64(ptr %ptr, double %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: v_dual_mov_b32 v5, v1 :: v_dual_mov_b32 v4, v0 ; GFX1250-NEXT: s_mov_b64 s[0:1], src_shared_base ; GFX1250-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 -; GFX1250-NEXT: v_cmpx_ne_u32_e64 s1, v1 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-NEXT: v_cmpx_ne_u32_e64 s1, v5 ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB35_6 -; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.check.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s1, v1 -; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 +; GFX1250-NEXT: s_cbranch_execnz .LBB35_3 +; GFX1250-NEXT: ; %bb.1: ; %Flow2 +; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 +; GFX1250-NEXT: s_cbranch_execnz .LBB35_8 +; GFX1250-NEXT: .LBB35_2: ; %atomicrmw.phi +; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 +; GFX1250-NEXT: s_wait_loadcnt 0x0 +; GFX1250-NEXT: s_set_pc_i64 s[30:31] +; GFX1250-NEXT: .LBB35_3: ; %atomicrmw.check.private +; GFX1250-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-NEXT: s_and_saveexec_b32 s1, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s1, exec_lo, s1 -; GFX1250-NEXT: s_cbranch_execz .LBB35_3 -; GFX1250-NEXT: ; %bb.2: ; %atomicrmw.global -; GFX1250-NEXT: global_atomic_add_f64 v[4:5], v[0:1], v[2:3], off th:TH_ATOMIC_RETURN scope:SCOPE_SYS -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_cbranch_execz .LBB35_5 +; GFX1250-NEXT: ; %bb.4: ; %atomicrmw.global +; GFX1250-NEXT: global_atomic_add_f64 v[0:1], v[4:5], v[2:3], off th:TH_ATOMIC_RETURN scope:SCOPE_SYS +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB35_3: ; %Flow +; GFX1250-NEXT: .LBB35_5: ; %Flow ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s1, s1 -; GFX1250-NEXT: s_cbranch_execz .LBB35_5 -; GFX1250-NEXT: ; %bb.4: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] +; GFX1250-NEXT: s_cbranch_execz .LBB35_7 +; GFX1250-NEXT: ; %bb.6: ; %atomicrmw.private +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s2, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo -; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off +; GFX1250-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo +; GFX1250-NEXT: scratch_load_b64 v[0:1], v4, off ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_add_f64_e32 v[0:1], v[4:5], v[2:3] -; GFX1250-NEXT: scratch_store_b64 v6, v[0:1], off -; GFX1250-NEXT: .LBB35_5: ; %Flow1 +; GFX1250-NEXT: v_add_f64_e32 v[2:3], v[0:1], v[2:3] +; GFX1250-NEXT: scratch_store_b64 v4, v[2:3], off +; GFX1250-NEXT: .LBB35_7: ; %Flow1 ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s1 -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB35_6: ; %Flow2 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB35_8 -; GFX1250-NEXT: ; %bb.7: ; %atomicrmw.shared -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-NEXT: v_cndmask_b32_e32 v0, -1, v0, vcc_lo +; GFX1250-NEXT: s_cbranch_execz .LBB35_2 +; GFX1250-NEXT: .LBB35_8: ; %atomicrmw.shared +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: ds_add_rtn_f64 v[4:5], v0, v[2:3] -; GFX1250-NEXT: .LBB35_8: ; %atomicrmw.phi +; GFX1250-NEXT: v_cndmask_b32_e32 v0, -1, v4, vcc_lo +; GFX1250-NEXT: ds_add_rtn_f64 v[0:1], v0, v[2:3] ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 -; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, v4 :: v_dual_mov_b32 v1, v5 +; GFX1250-NEXT: s_wait_dscnt 0x0 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %result = atomicrmw fadd ptr %ptr, double %val syncscope("one-as") monotonic ret double %result @@ -686,40 +692,42 @@ define double @flat_system_atomic_fmin_f64(ptr %ptr, double %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 +; GFX1250-NEXT: v_dual_mov_b32 v5, v1 :: v_dual_mov_b32 v4, v0 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 +; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB38_2 -; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global -; GFX1250-NEXT: flat_atomic_min_num_f64 v[4:5], v[0:1], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_cbranch_execnz .LBB38_3 +; GFX1250-NEXT: ; %bb.1: ; %Flow +; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 +; GFX1250-NEXT: s_cbranch_execnz .LBB38_4 +; GFX1250-NEXT: .LBB38_2: ; %atomicrmw.phi +; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_set_pc_i64 s[30:31] +; GFX1250-NEXT: .LBB38_3: ; %atomicrmw.global +; GFX1250-NEXT: flat_atomic_min_num_f64 v[0:1], v[4:5], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB38_2: ; %Flow ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB38_4 -; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] +; GFX1250-NEXT: s_cbranch_execz .LBB38_2 +; GFX1250-NEXT: .LBB38_4: ; %atomicrmw.private +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v4, vcc_lo -; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off +; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v0, vcc_lo +; GFX1250-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[4:5], v[4:5] -; GFX1250-NEXT: v_min_num_f64_e32 v[0:1], v[0:1], v[2:3] -; GFX1250-NEXT: scratch_store_b64 v6, v[0:1], off -; GFX1250-NEXT: .LBB38_4: ; %atomicrmw.phi +; GFX1250-NEXT: v_max_num_f64_e32 v[4:5], v[0:1], v[0:1] +; GFX1250-NEXT: v_min_num_f64_e32 v[2:3], v[4:5], v[2:3] +; GFX1250-NEXT: scratch_store_b64 v6, v[2:3], off ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 -; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, v4 :: v_dual_mov_b32 v1, v5 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %result = atomicrmw fmin ptr %ptr, double %val monotonic ret double %result @@ -730,40 +738,42 @@ define double @flat_one_as_atomic_fmin_f64(ptr %ptr, double %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 +; GFX1250-NEXT: v_dual_mov_b32 v5, v1 :: v_dual_mov_b32 v4, v0 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 +; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB39_2 -; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global -; GFX1250-NEXT: flat_atomic_min_num_f64 v[4:5], v[0:1], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_cbranch_execnz .LBB39_3 +; GFX1250-NEXT: ; %bb.1: ; %Flow +; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 +; GFX1250-NEXT: s_cbranch_execnz .LBB39_4 +; GFX1250-NEXT: .LBB39_2: ; %atomicrmw.phi +; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_set_pc_i64 s[30:31] +; GFX1250-NEXT: .LBB39_3: ; %atomicrmw.global +; GFX1250-NEXT: flat_atomic_min_num_f64 v[0:1], v[4:5], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB39_2: ; %Flow ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB39_4 -; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] +; GFX1250-NEXT: s_cbranch_execz .LBB39_2 +; GFX1250-NEXT: .LBB39_4: ; %atomicrmw.private +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v4, vcc_lo -; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off +; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v0, vcc_lo +; GFX1250-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[4:5], v[4:5] -; GFX1250-NEXT: v_min_num_f64_e32 v[0:1], v[0:1], v[2:3] -; GFX1250-NEXT: scratch_store_b64 v6, v[0:1], off -; GFX1250-NEXT: .LBB39_4: ; %atomicrmw.phi +; GFX1250-NEXT: v_max_num_f64_e32 v[4:5], v[0:1], v[0:1] +; GFX1250-NEXT: v_min_num_f64_e32 v[2:3], v[4:5], v[2:3] +; GFX1250-NEXT: scratch_store_b64 v6, v[2:3], off ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 -; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, v4 :: v_dual_mov_b32 v1, v5 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %result = atomicrmw fmin ptr %ptr, double %val syncscope("one-as") monotonic ret double %result @@ -798,40 +808,42 @@ define double @flat_system_atomic_fmax_f64(ptr %ptr, double %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 +; GFX1250-NEXT: v_dual_mov_b32 v5, v1 :: v_dual_mov_b32 v4, v0 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 +; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB42_2 -; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global -; GFX1250-NEXT: flat_atomic_max_num_f64 v[4:5], v[0:1], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_cbranch_execnz .LBB42_3 +; GFX1250-NEXT: ; %bb.1: ; %Flow +; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 +; GFX1250-NEXT: s_cbranch_execnz .LBB42_4 +; GFX1250-NEXT: .LBB42_2: ; %atomicrmw.phi +; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_set_pc_i64 s[30:31] +; GFX1250-NEXT: .LBB42_3: ; %atomicrmw.global +; GFX1250-NEXT: flat_atomic_max_num_f64 v[0:1], v[4:5], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB42_2: ; %Flow ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB42_4 -; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] +; GFX1250-NEXT: s_cbranch_execz .LBB42_2 +; GFX1250-NEXT: .LBB42_4: ; %atomicrmw.private +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v4, vcc_lo -; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off +; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v0, vcc_lo +; GFX1250-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[4:5], v[4:5] -; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[2:3] -; GFX1250-NEXT: scratch_store_b64 v6, v[0:1], off -; GFX1250-NEXT: .LBB42_4: ; %atomicrmw.phi +; GFX1250-NEXT: v_max_num_f64_e32 v[4:5], v[0:1], v[0:1] +; GFX1250-NEXT: v_max_num_f64_e32 v[2:3], v[4:5], v[2:3] +; GFX1250-NEXT: scratch_store_b64 v6, v[2:3], off ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 -; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, v4 :: v_dual_mov_b32 v1, v5 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %result = atomicrmw fmax ptr %ptr, double %val monotonic ret double %result @@ -842,40 +854,42 @@ define double @flat_one_as_atomic_fmax_f64(ptr %ptr, double %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 -; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 +; GFX1250-NEXT: v_dual_mov_b32 v5, v1 :: v_dual_mov_b32 v4, v0 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 +; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 +; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB43_2 -; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global -; GFX1250-NEXT: flat_atomic_max_num_f64 v[4:5], v[0:1], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS -; GFX1250-NEXT: ; implicit-def: $vgpr0_vgpr1 +; GFX1250-NEXT: s_cbranch_execnz .LBB43_3 +; GFX1250-NEXT: ; %bb.1: ; %Flow +; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 +; GFX1250-NEXT: s_cbranch_execnz .LBB43_4 +; GFX1250-NEXT: .LBB43_2: ; %atomicrmw.phi +; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_set_pc_i64 s[30:31] +; GFX1250-NEXT: .LBB43_3: ; %atomicrmw.global +; GFX1250-NEXT: flat_atomic_max_num_f64 v[0:1], v[4:5], v[2:3] th:TH_ATOMIC_RETURN scope:SCOPE_SYS +; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-NEXT: .LBB43_2: ; %Flow ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 -; GFX1250-NEXT: s_cbranch_execz .LBB43_4 -; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo -; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] +; GFX1250-NEXT: s_cbranch_execz .LBB43_2 +; GFX1250-NEXT: .LBB43_4: ; %atomicrmw.private +; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v4, vcc_lo -; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off +; GFX1250-NEXT: v_dual_max_num_f64 v[2:3], v[2:3], v[2:3] :: v_dual_cndmask_b32 v6, -1, v0, vcc_lo +; GFX1250-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[4:5], v[4:5] -; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[2:3] -; GFX1250-NEXT: scratch_store_b64 v6, v[0:1], off -; GFX1250-NEXT: .LBB43_4: ; %atomicrmw.phi +; GFX1250-NEXT: v_max_num_f64_e32 v[4:5], v[0:1], v[0:1] +; GFX1250-NEXT: v_max_num_f64_e32 v[2:3], v[4:5], v[2:3] +; GFX1250-NEXT: scratch_store_b64 v6, v[2:3], off ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: s_or_b32 exec_lo, exec_lo, s0 -; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, v4 :: v_dual_mov_b32 v1, v5 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %result = atomicrmw fmax ptr %ptr, double %val syncscope("one-as") monotonic ret double %result @@ -982,13 +996,11 @@ define i64 @flat_one_as_atomic_min_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB52_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1000,10 +1012,9 @@ define i64 @flat_one_as_atomic_min_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB52_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off @@ -1025,13 +1036,11 @@ define i64 @flat_system_atomic_min_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB53_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1043,10 +1052,9 @@ define i64 @flat_system_atomic_min_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB53_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off @@ -1068,13 +1076,11 @@ define i64 @flat_one_as_atomic_max_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB54_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1086,10 +1092,9 @@ define i64 @flat_one_as_atomic_max_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB54_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off @@ -1111,13 +1116,11 @@ define i64 @flat_system_atomic_max_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB55_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1129,10 +1132,9 @@ define i64 @flat_system_atomic_max_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB55_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off @@ -1154,13 +1156,11 @@ define i64 @flat_one_as_atomic_umin_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB56_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1172,10 +1172,9 @@ define i64 @flat_one_as_atomic_umin_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB56_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off @@ -1197,13 +1196,11 @@ define i64 @flat_system_atomic_umin_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB57_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1215,10 +1212,9 @@ define i64 @flat_system_atomic_umin_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB57_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off @@ -1240,13 +1236,11 @@ define i64 @flat_one_as_atomic_umax_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB58_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1258,10 +1252,9 @@ define i64 @flat_one_as_atomic_umax_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB58_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off @@ -1283,13 +1276,11 @@ define i64 @flat_system_atomic_umax_i64(ptr %ptr, i64 %val) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-NEXT: v_xor_b32_e32 v4, s0, v1 +; GFX1250-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v4 ; GFX1250-NEXT: ; implicit-def: $vgpr4_vgpr5 ; GFX1250-NEXT: s_and_saveexec_b32 s0, vcc_lo -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB59_2 ; GFX1250-NEXT: ; %bb.1: ; %atomicrmw.global @@ -1301,10 +1292,9 @@ define i64 @flat_system_atomic_umax_i64(ptr %ptr, i64 %val) { ; GFX1250-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-NEXT: s_cbranch_execz .LBB59_4 ; GFX1250-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, s1, v0 +; GFX1250-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-NEXT: scratch_load_b64 v[4:5], v6, off diff --git a/llvm/test/CodeGen/AMDGPU/bf16.ll b/llvm/test/CodeGen/AMDGPU/bf16.ll index 94ba5cd..6b5647e 100644 --- a/llvm/test/CodeGen/AMDGPU/bf16.ll +++ b/llvm/test/CodeGen/AMDGPU/bf16.ll @@ -569,10 +569,10 @@ define <16 x bfloat> @v_load_global_v16bf16(ptr addrspace(1) %ptr) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v5, v1 :: v_dual_mov_b32 v4, v0 +; GFX1250-NEXT: v_dual_mov_b32 v9, v1 :: v_dual_mov_b32 v8, v0 ; GFX1250-NEXT: s_clause 0x1 -; GFX1250-NEXT: global_load_b128 v[0:3], v[4:5], off -; GFX1250-NEXT: global_load_b128 v[4:7], v[4:5], off offset:16 +; GFX1250-NEXT: global_load_b128 v[0:3], v[8:9], off +; GFX1250-NEXT: global_load_b128 v[4:7], v[8:9], off offset:16 ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %load = load <16 x bfloat>, ptr addrspace(1) %ptr @@ -752,12 +752,12 @@ define <32 x bfloat> @v_load_global_v32bf16(ptr addrspace(1) %ptr) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v13, v1 :: v_dual_mov_b32 v12, v0 +; GFX1250-NEXT: v_dual_mov_b32 v17, v1 :: v_dual_mov_b32 v16, v0 ; GFX1250-NEXT: s_clause 0x3 -; GFX1250-NEXT: global_load_b128 v[0:3], v[12:13], off -; GFX1250-NEXT: global_load_b128 v[4:7], v[12:13], off offset:16 -; GFX1250-NEXT: global_load_b128 v[8:11], v[12:13], off offset:32 -; GFX1250-NEXT: global_load_b128 v[12:15], v[12:13], off offset:48 +; GFX1250-NEXT: global_load_b128 v[0:3], v[16:17], off +; GFX1250-NEXT: global_load_b128 v[4:7], v[16:17], off offset:16 +; GFX1250-NEXT: global_load_b128 v[8:11], v[16:17], off offset:32 +; GFX1250-NEXT: global_load_b128 v[12:15], v[16:17], off offset:48 ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %load = load <32 x bfloat>, ptr addrspace(1) %ptr @@ -1055,16 +1055,16 @@ define <64 x bfloat> @v_load_global_v64bf16(ptr addrspace(1) %ptr) { ; GFX1250: ; %bb.0: ; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v29, v1 :: v_dual_mov_b32 v28, v0 +; GFX1250-NEXT: v_dual_mov_b32 v33, v1 :: v_dual_mov_b32 v32, v0 ; GFX1250-NEXT: s_clause 0x7 -; GFX1250-NEXT: global_load_b128 v[0:3], v[28:29], off -; GFX1250-NEXT: global_load_b128 v[4:7], v[28:29], off offset:16 -; GFX1250-NEXT: global_load_b128 v[8:11], v[28:29], off offset:32 -; GFX1250-NEXT: global_load_b128 v[12:15], v[28:29], off offset:48 -; GFX1250-NEXT: global_load_b128 v[16:19], v[28:29], off offset:64 -; GFX1250-NEXT: global_load_b128 v[20:23], v[28:29], off offset:80 -; GFX1250-NEXT: global_load_b128 v[24:27], v[28:29], off offset:96 -; GFX1250-NEXT: global_load_b128 v[28:31], v[28:29], off offset:112 +; GFX1250-NEXT: global_load_b128 v[0:3], v[32:33], off +; GFX1250-NEXT: global_load_b128 v[4:7], v[32:33], off offset:16 +; GFX1250-NEXT: global_load_b128 v[8:11], v[32:33], off offset:32 +; GFX1250-NEXT: global_load_b128 v[12:15], v[32:33], off offset:48 +; GFX1250-NEXT: global_load_b128 v[16:19], v[32:33], off offset:64 +; GFX1250-NEXT: global_load_b128 v[20:23], v[32:33], off offset:80 +; GFX1250-NEXT: global_load_b128 v[24:27], v[32:33], off offset:96 +; GFX1250-NEXT: global_load_b128 v[28:31], v[32:33], off offset:112 ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: s_set_pc_i64 s[30:31] %load = load <64 x bfloat>, ptr addrspace(1) %ptr diff --git a/llvm/test/CodeGen/AMDGPU/calling-conventions.ll b/llvm/test/CodeGen/AMDGPU/calling-conventions.ll index ddd3b152..363a248 100644 --- a/llvm/test/CodeGen/AMDGPU/calling-conventions.ll +++ b/llvm/test/CodeGen/AMDGPU/calling-conventions.ll @@ -2700,142 +2700,142 @@ define amdgpu_kernel void @amd_kernel_v32i8(<32 x i8> %arg0) { ; ; GFX1250-LABEL: amd_kernel_v32i8: ; GFX1250: ; %bb.0: ; %entry -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 ; GFX1250-NEXT: v_mov_b64_e32 v[8:9], 16 ; GFX1250-NEXT: v_mov_b64_e32 v[10:11], 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_lshr_b32 s16, s0, 16 -; GFX1250-NEXT: s_lshr_b32 s17, s0, 24 -; GFX1250-NEXT: s_lshr_b32 s20, s2, 16 -; GFX1250-NEXT: s_lshr_b32 s21, s2, 24 -; GFX1250-NEXT: s_lshr_b32 s14, s7, 16 -; GFX1250-NEXT: s_lshr_b32 s15, s7, 24 -; GFX1250-NEXT: s_bfe_u32 s27, s7, 0x80008 +; GFX1250-NEXT: s_lshr_b32 s16, s8, 16 +; GFX1250-NEXT: s_lshr_b32 s17, s8, 24 +; GFX1250-NEXT: s_lshr_b32 s6, s15, 16 +; GFX1250-NEXT: s_lshr_b32 s7, s15, 24 +; GFX1250-NEXT: s_bfe_u32 s27, s15, 0x80008 ; GFX1250-NEXT: s_add_co_i32 s17, s17, s17 ; GFX1250-NEXT: s_add_co_i32 s16, s16, s16 -; GFX1250-NEXT: s_lshr_b32 s18, s1, 16 -; GFX1250-NEXT: s_lshr_b32 s19, s1, 24 -; GFX1250-NEXT: s_lshr_b32 s22, s3, 16 -; GFX1250-NEXT: s_lshr_b32 s23, s3, 24 -; GFX1250-NEXT: s_bfe_u32 s29, s1, 0x80008 -; GFX1250-NEXT: s_bfe_u32 s30, s3, 0x80008 -; GFX1250-NEXT: s_add_co_i32 s21, s21, s21 -; GFX1250-NEXT: s_add_co_i32 s20, s20, s20 ; GFX1250-NEXT: s_lshl_b32 s17, s17, 8 ; GFX1250-NEXT: s_and_b32 s16, s16, 0xff -; GFX1250-NEXT: s_add_co_i32 s7, s7, s7 -; GFX1250-NEXT: s_add_co_i32 s27, s27, s27 ; GFX1250-NEXT: s_add_co_i32 s15, s15, s15 -; GFX1250-NEXT: s_add_co_i32 s14, s14, s14 -; GFX1250-NEXT: s_add_co_i32 s3, s3, s3 +; GFX1250-NEXT: s_add_co_i32 s27, s27, s27 +; GFX1250-NEXT: s_add_co_i32 s7, s7, s7 +; GFX1250-NEXT: s_add_co_i32 s6, s6, s6 +; GFX1250-NEXT: s_or_b32 s16, s16, s17 +; GFX1250-NEXT: s_and_b32 s15, s15, 0xff +; GFX1250-NEXT: s_lshl_b32 s17, s27, 8 +; GFX1250-NEXT: s_lshl_b32 s7, s7, 8 +; GFX1250-NEXT: s_and_b32 s6, s6, 0xff +; GFX1250-NEXT: s_or_b32 s15, s15, s17 +; GFX1250-NEXT: s_or_b32 s6, s6, s7 +; GFX1250-NEXT: s_bfe_u32 s26, s14, 0x80008 +; GFX1250-NEXT: s_and_b32 s7, s15, 0xffff +; GFX1250-NEXT: s_lshl_b32 s6, s6, 16 +; GFX1250-NEXT: s_lshr_b32 s20, s10, 16 +; GFX1250-NEXT: s_lshr_b32 s21, s10, 24 +; GFX1250-NEXT: s_lshr_b32 s4, s14, 16 +; GFX1250-NEXT: s_lshr_b32 s5, s14, 24 +; GFX1250-NEXT: s_or_b32 s6, s7, s6 +; GFX1250-NEXT: s_add_co_i32 s7, s14, s14 +; GFX1250-NEXT: s_add_co_i32 s26, s26, s26 +; GFX1250-NEXT: s_lshr_b32 s18, s9, 16 +; GFX1250-NEXT: s_lshr_b32 s19, s9, 24 +; GFX1250-NEXT: s_lshr_b32 s22, s11, 16 +; GFX1250-NEXT: s_lshr_b32 s23, s11, 24 +; GFX1250-NEXT: s_bfe_u32 s29, s9, 0x80008 +; GFX1250-NEXT: s_bfe_u32 s30, s11, 0x80008 +; GFX1250-NEXT: s_add_co_i32 s21, s21, s21 +; GFX1250-NEXT: s_add_co_i32 s20, s20, s20 +; GFX1250-NEXT: s_lshr_b32 s2, s13, 16 +; GFX1250-NEXT: s_lshr_b32 s3, s13, 24 +; GFX1250-NEXT: s_and_b32 s7, s7, 0xff +; GFX1250-NEXT: s_lshl_b32 s14, s26, 8 +; GFX1250-NEXT: s_add_co_i32 s5, s5, s5 +; GFX1250-NEXT: s_add_co_i32 s4, s4, s4 +; GFX1250-NEXT: s_add_co_i32 s11, s11, s11 ; GFX1250-NEXT: s_add_co_i32 s30, s30, s30 ; GFX1250-NEXT: s_add_co_i32 s23, s23, s23 ; GFX1250-NEXT: s_add_co_i32 s22, s22, s22 ; GFX1250-NEXT: s_lshl_b32 s21, s21, 8 ; GFX1250-NEXT: s_and_b32 s20, s20, 0xff -; GFX1250-NEXT: s_add_co_i32 s1, s1, s1 +; GFX1250-NEXT: s_add_co_i32 s9, s9, s9 ; GFX1250-NEXT: s_add_co_i32 s29, s29, s29 ; GFX1250-NEXT: s_add_co_i32 s19, s19, s19 ; GFX1250-NEXT: s_add_co_i32 s18, s18, s18 -; GFX1250-NEXT: s_lshr_b32 s10, s5, 16 -; GFX1250-NEXT: s_lshr_b32 s11, s5, 24 -; GFX1250-NEXT: s_lshr_b32 s12, s6, 16 -; GFX1250-NEXT: s_lshr_b32 s13, s6, 24 -; GFX1250-NEXT: s_or_b32 s16, s16, s17 -; GFX1250-NEXT: s_and_b32 s7, s7, 0xff -; GFX1250-NEXT: s_lshl_b32 s17, s27, 8 -; GFX1250-NEXT: s_lshl_b32 s15, s15, 8 -; GFX1250-NEXT: s_and_b32 s14, s14, 0xff -; GFX1250-NEXT: s_and_b32 s3, s3, 0xff +; GFX1250-NEXT: s_bfe_u32 s25, s13, 0x80008 +; GFX1250-NEXT: s_lshl_b32 s5, s5, 8 +; GFX1250-NEXT: s_and_b32 s4, s4, 0xff +; GFX1250-NEXT: s_or_b32 s7, s7, s14 +; GFX1250-NEXT: s_add_co_i32 s3, s3, s3 +; GFX1250-NEXT: s_add_co_i32 s2, s2, s2 +; GFX1250-NEXT: s_and_b32 s11, s11, 0xff ; GFX1250-NEXT: s_lshl_b32 s30, s30, 8 ; GFX1250-NEXT: s_lshl_b32 s23, s23, 8 ; GFX1250-NEXT: s_and_b32 s22, s22, 0xff ; GFX1250-NEXT: s_or_b32 s20, s20, s21 -; GFX1250-NEXT: s_and_b32 s1, s1, 0xff +; GFX1250-NEXT: s_and_b32 s9, s9, 0xff ; GFX1250-NEXT: s_lshl_b32 s21, s29, 8 ; GFX1250-NEXT: s_lshl_b32 s19, s19, 8 ; GFX1250-NEXT: s_and_b32 s18, s18, 0xff -; GFX1250-NEXT: s_lshr_b32 s8, s4, 16 -; GFX1250-NEXT: s_lshr_b32 s9, s4, 24 -; GFX1250-NEXT: s_bfe_u32 s24, s4, 0x80008 -; GFX1250-NEXT: s_bfe_u32 s25, s5, 0x80008 -; GFX1250-NEXT: s_bfe_u32 s26, s6, 0x80008 -; GFX1250-NEXT: s_or_b32 s7, s7, s17 -; GFX1250-NEXT: s_or_b32 s14, s14, s15 -; GFX1250-NEXT: s_add_co_i32 s13, s13, s13 -; GFX1250-NEXT: s_add_co_i32 s12, s12, s12 -; GFX1250-NEXT: s_add_co_i32 s11, s11, s11 -; GFX1250-NEXT: s_add_co_i32 s10, s10, s10 -; GFX1250-NEXT: s_bfe_u32 s28, s0, 0x80008 -; GFX1250-NEXT: s_or_b32 s3, s3, s30 +; GFX1250-NEXT: s_lshr_b32 s0, s12, 16 +; GFX1250-NEXT: s_lshr_b32 s1, s12, 24 +; GFX1250-NEXT: s_bfe_u32 s24, s12, 0x80008 +; GFX1250-NEXT: s_or_b32 s4, s4, s5 +; GFX1250-NEXT: s_and_b32 s5, s7, 0xffff +; GFX1250-NEXT: s_add_co_i32 s7, s13, s13 +; GFX1250-NEXT: s_add_co_i32 s25, s25, s25 +; GFX1250-NEXT: s_lshl_b32 s3, s3, 8 +; GFX1250-NEXT: s_and_b32 s2, s2, 0xff +; GFX1250-NEXT: s_bfe_u32 s28, s8, 0x80008 +; GFX1250-NEXT: s_or_b32 s11, s11, s30 ; GFX1250-NEXT: s_or_b32 s22, s22, s23 -; GFX1250-NEXT: s_bfe_u32 s23, s2, 0x80008 -; GFX1250-NEXT: s_or_b32 s1, s1, s21 +; GFX1250-NEXT: s_bfe_u32 s23, s10, 0x80008 +; GFX1250-NEXT: s_or_b32 s9, s9, s21 ; GFX1250-NEXT: s_or_b32 s18, s18, s19 -; GFX1250-NEXT: s_and_b32 s7, s7, 0xffff -; GFX1250-NEXT: s_lshl_b32 s14, s14, 16 -; GFX1250-NEXT: s_add_co_i32 s6, s6, s6 -; GFX1250-NEXT: s_add_co_i32 s26, s26, s26 -; GFX1250-NEXT: s_lshl_b32 s13, s13, 8 -; GFX1250-NEXT: s_and_b32 s12, s12, 0xff -; GFX1250-NEXT: s_add_co_i32 s5, s5, s5 -; GFX1250-NEXT: s_add_co_i32 s25, s25, s25 -; GFX1250-NEXT: s_lshl_b32 s11, s11, 8 -; GFX1250-NEXT: s_and_b32 s10, s10, 0xff -; GFX1250-NEXT: s_add_co_i32 s4, s4, s4 +; GFX1250-NEXT: s_lshl_b32 s4, s4, 16 +; GFX1250-NEXT: s_and_b32 s7, s7, 0xff +; GFX1250-NEXT: s_lshl_b32 s13, s25, 8 +; GFX1250-NEXT: s_or_b32 s2, s2, s3 +; GFX1250-NEXT: s_add_co_i32 s3, s12, s12 ; GFX1250-NEXT: s_add_co_i32 s24, s24, s24 -; GFX1250-NEXT: s_add_co_i32 s9, s9, s9 -; GFX1250-NEXT: s_add_co_i32 s8, s8, s8 -; GFX1250-NEXT: s_and_b32 s3, s3, 0xffff +; GFX1250-NEXT: s_add_co_i32 s1, s1, s1 +; GFX1250-NEXT: s_add_co_i32 s0, s0, s0 +; GFX1250-NEXT: s_and_b32 s11, s11, 0xffff ; GFX1250-NEXT: s_lshl_b32 s22, s22, 16 -; GFX1250-NEXT: s_add_co_i32 s2, s2, s2 +; GFX1250-NEXT: s_add_co_i32 s10, s10, s10 ; GFX1250-NEXT: s_add_co_i32 s23, s23, s23 -; GFX1250-NEXT: s_and_b32 s1, s1, 0xffff +; GFX1250-NEXT: s_and_b32 s9, s9, 0xffff ; GFX1250-NEXT: s_lshl_b32 s18, s18, 16 -; GFX1250-NEXT: s_add_co_i32 s0, s0, s0 +; GFX1250-NEXT: s_add_co_i32 s8, s8, s8 ; GFX1250-NEXT: s_add_co_i32 s28, s28, s28 -; GFX1250-NEXT: s_or_b32 s7, s7, s14 -; GFX1250-NEXT: s_and_b32 s6, s6, 0xff -; GFX1250-NEXT: s_lshl_b32 s14, s26, 8 -; GFX1250-NEXT: s_or_b32 s12, s12, s13 -; GFX1250-NEXT: s_and_b32 s5, s5, 0xff -; GFX1250-NEXT: s_lshl_b32 s13, s25, 8 -; GFX1250-NEXT: s_or_b32 s10, s10, s11 -; GFX1250-NEXT: s_and_b32 s4, s4, 0xff -; GFX1250-NEXT: s_lshl_b32 s11, s24, 8 -; GFX1250-NEXT: s_lshl_b32 s9, s9, 8 -; GFX1250-NEXT: s_and_b32 s8, s8, 0xff -; GFX1250-NEXT: s_or_b32 s3, s3, s22 -; GFX1250-NEXT: s_and_b32 s2, s2, 0xff -; GFX1250-NEXT: s_lshl_b32 s22, s23, 8 -; GFX1250-NEXT: s_or_b32 s1, s1, s18 +; GFX1250-NEXT: s_or_b32 s4, s5, s4 +; GFX1250-NEXT: s_or_b32 s5, s7, s13 +; GFX1250-NEXT: s_and_b32 s3, s3, 0xff +; GFX1250-NEXT: s_lshl_b32 s7, s24, 8 +; GFX1250-NEXT: s_lshl_b32 s1, s1, 8 ; GFX1250-NEXT: s_and_b32 s0, s0, 0xff +; GFX1250-NEXT: s_or_b32 s11, s11, s22 +; GFX1250-NEXT: s_and_b32 s10, s10, 0xff +; GFX1250-NEXT: s_lshl_b32 s22, s23, 8 +; GFX1250-NEXT: s_or_b32 s9, s9, s18 +; GFX1250-NEXT: s_and_b32 s8, s8, 0xff ; GFX1250-NEXT: s_lshl_b32 s18, s28, 8 -; GFX1250-NEXT: s_or_b32 s6, s6, s14 -; GFX1250-NEXT: s_or_b32 s5, s5, s13 -; GFX1250-NEXT: s_or_b32 s4, s4, s11 -; GFX1250-NEXT: s_or_b32 s8, s8, s9 -; GFX1250-NEXT: s_or_b32 s2, s2, s22 -; GFX1250-NEXT: s_or_b32 s0, s0, s18 -; GFX1250-NEXT: s_and_b32 s6, s6, 0xffff -; GFX1250-NEXT: s_lshl_b32 s12, s12, 16 +; GFX1250-NEXT: s_or_b32 s3, s3, s7 +; GFX1250-NEXT: s_or_b32 s0, s0, s1 +; GFX1250-NEXT: s_or_b32 s10, s10, s22 +; GFX1250-NEXT: s_or_b32 s8, s8, s18 ; GFX1250-NEXT: s_and_b32 s5, s5, 0xffff -; GFX1250-NEXT: s_and_b32 s4, s4, 0xffff -; GFX1250-NEXT: s_lshl_b32 s8, s8, 16 -; GFX1250-NEXT: s_lshl_b32 s9, s10, 16 -; GFX1250-NEXT: s_and_b32 s2, s2, 0xffff +; GFX1250-NEXT: s_and_b32 s1, s3, 0xffff +; GFX1250-NEXT: s_lshl_b32 s0, s0, 16 +; GFX1250-NEXT: s_lshl_b32 s2, s2, 16 +; GFX1250-NEXT: s_and_b32 s10, s10, 0xffff ; GFX1250-NEXT: s_lshl_b32 s20, s20, 16 -; GFX1250-NEXT: s_and_b32 s0, s0, 0xffff +; GFX1250-NEXT: s_and_b32 s8, s8, 0xffff ; GFX1250-NEXT: s_lshl_b32 s16, s16, 16 -; GFX1250-NEXT: s_or_b32 s6, s6, s12 -; GFX1250-NEXT: s_or_b32 s4, s4, s8 -; GFX1250-NEXT: s_or_b32 s5, s5, s9 -; GFX1250-NEXT: s_or_b32 s2, s2, s20 -; GFX1250-NEXT: s_or_b32 s0, s0, s16 -; GFX1250-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 -; GFX1250-NEXT: v_dual_mov_b32 v2, s6 :: v_dual_mov_b32 v3, s7 -; GFX1250-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, s1 -; GFX1250-NEXT: v_dual_mov_b32 v6, s2 :: v_dual_mov_b32 v7, s3 +; GFX1250-NEXT: s_or_b32 s0, s1, s0 +; GFX1250-NEXT: s_or_b32 s1, s5, s2 +; GFX1250-NEXT: s_or_b32 s10, s10, s20 +; GFX1250-NEXT: s_or_b32 s8, s8, s16 +; GFX1250-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 +; GFX1250-NEXT: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s6 +; GFX1250-NEXT: v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, s9 +; GFX1250-NEXT: v_dual_mov_b32 v6, s10 :: v_dual_mov_b32 v7, s11 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_store_b128 v[8:9], v[0:3], off ; GFX1250-NEXT: global_store_b128 v[10:11], v[4:7], off diff --git a/llvm/test/CodeGen/AMDGPU/carryout-selection.ll b/llvm/test/CodeGen/AMDGPU/carryout-selection.ll index 51652a0..2ae6fc2 100644 --- a/llvm/test/CodeGen/AMDGPU/carryout-selection.ll +++ b/llvm/test/CodeGen/AMDGPU/carryout-selection.ll @@ -117,12 +117,12 @@ define amdgpu_kernel void @sadd64rr(ptr addrspace(1) %out, i64 %a, i64 %b) { ; ; GFX1250-LABEL: sadd64rr: ; GFX1250: ; %bb.0: ; %entry +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_add_nc_u64 s[2:3], s[2:3], s[4:5] +; GFX1250-NEXT: s_add_nc_u64 s[2:3], s[2:3], s[6:7] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] @@ -818,17 +818,17 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car ; ; GFX1250-LABEL: suaddo64: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_add_nc_u64 s[6:7], s[4:5], s[6:7] -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) -; GFX1250-NEXT: v_cmp_lt_u64_e64 s4, s[6:7], s[4:5] -; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[6:7] -; GFX1250-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4 +; GFX1250-NEXT: s_add_nc_u64 s[0:1], s[12:13], s[14:15] +; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) +; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[0:1] +; GFX1250-NEXT: v_cmp_lt_u64_e64 s0, s[0:1], s[12:13] +; GFX1250-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0 ; GFX1250-NEXT: s_clause 0x1 -; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] -; GFX1250-NEXT: global_store_b8 v2, v3, s[2:3] +; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[8:9] +; GFX1250-NEXT: global_store_b8 v2, v3, s[10:11] ; GFX1250-NEXT: s_endpgm %uadd = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 %a, i64 %b) %val = extractvalue { i64, i1 } %uadd, 0 @@ -1096,12 +1096,12 @@ define amdgpu_kernel void @ssub64rr(ptr addrspace(1) %out, i64 %a, i64 %b) { ; ; GFX1250-LABEL: ssub64rr: ; GFX1250: ; %bb.0: ; %entry +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_sub_nc_u64 s[2:3], s[2:3], s[4:5] +; GFX1250-NEXT: s_sub_nc_u64 s[2:3], s[2:3], s[6:7] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] @@ -1798,17 +1798,17 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car ; ; GFX1250-LABEL: susubo64: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_sub_nc_u64 s[6:7], s[4:5], s[6:7] -; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) -; GFX1250-NEXT: v_cmp_gt_u64_e64 s4, s[6:7], s[4:5] -; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[6:7] -; GFX1250-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4 +; GFX1250-NEXT: s_sub_nc_u64 s[0:1], s[12:13], s[14:15] +; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) +; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[0:1] +; GFX1250-NEXT: v_cmp_gt_u64_e64 s0, s[0:1], s[12:13] +; GFX1250-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0 ; GFX1250-NEXT: s_clause 0x1 -; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] -; GFX1250-NEXT: global_store_b8 v2, v3, s[2:3] +; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[8:9] +; GFX1250-NEXT: global_store_b8 v2, v3, s[10:11] ; GFX1250-NEXT: s_endpgm %usub = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 %a, i64 %b) %val = extractvalue { i64, i1 } %usub, 0 @@ -3099,70 +3099,70 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) { ; ; GFX1250-LABEL: sudiv64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_or_b64 s[6:7], s[2:3], s[4:5] +; GFX1250-NEXT: s_or_b64 s[4:5], s[2:3], s[6:7] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: s_and_b64 s[6:7], s[6:7], 0xffffffff00000000 -; GFX1250-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1250-NEXT: s_and_b64 s[4:5], s[4:5], 0xffffffff00000000 +; GFX1250-NEXT: s_cmp_lg_u64 s[4:5], 0 ; GFX1250-NEXT: s_cbranch_scc0 .LBB16_4 ; GFX1250-NEXT: ; %bb.1: -; GFX1250-NEXT: s_cvt_f32_u32 s6, s4 -; GFX1250-NEXT: s_cvt_f32_u32 s7, s5 -; GFX1250-NEXT: s_sub_nc_u64 s[10:11], 0, s[4:5] +; GFX1250-NEXT: s_cvt_f32_u32 s4, s6 +; GFX1250-NEXT: s_cvt_f32_u32 s5, s7 +; GFX1250-NEXT: s_sub_nc_u64 s[10:11], 0, s[6:7] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_2) | instskip(NEXT) | instid1(SALU_CYCLE_3) -; GFX1250-NEXT: s_fmac_f32 s6, s7, 0x4f800000 -; GFX1250-NEXT: v_s_rcp_f32 s6, s6 +; GFX1250-NEXT: s_fmac_f32 s4, s5, 0x4f800000 +; GFX1250-NEXT: v_s_rcp_f32 s4, s4 ; GFX1250-NEXT: s_delay_alu instid0(TRANS32_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_3) -; GFX1250-NEXT: s_mul_f32 s6, s6, 0x5f7ffffc -; GFX1250-NEXT: s_mul_f32 s7, s6, 0x2f800000 +; GFX1250-NEXT: s_mul_f32 s4, s4, 0x5f7ffffc +; GFX1250-NEXT: s_mul_f32 s5, s4, 0x2f800000 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_3) | instskip(NEXT) | instid1(SALU_CYCLE_3) -; GFX1250-NEXT: s_trunc_f32 s7, s7 -; GFX1250-NEXT: s_fmac_f32 s6, s7, 0xcf800000 -; GFX1250-NEXT: s_cvt_u32_f32 s9, s7 -; GFX1250-NEXT: s_mov_b32 s7, 0 +; GFX1250-NEXT: s_trunc_f32 s5, s5 +; GFX1250-NEXT: s_fmac_f32 s4, s5, 0xcf800000 +; GFX1250-NEXT: s_cvt_u32_f32 s9, s5 +; GFX1250-NEXT: s_mov_b32 s5, 0 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_3) -; GFX1250-NEXT: s_cvt_u32_f32 s8, s6 +; GFX1250-NEXT: s_cvt_u32_f32 s8, s4 ; GFX1250-NEXT: s_mul_u64 s[12:13], s[10:11], s[8:9] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_mul_hi_u32 s15, s8, s13 ; GFX1250-NEXT: s_mul_i32 s14, s8, s13 -; GFX1250-NEXT: s_mul_hi_u32 s6, s8, s12 +; GFX1250-NEXT: s_mul_hi_u32 s4, s8, s12 ; GFX1250-NEXT: s_mul_i32 s17, s9, s12 -; GFX1250-NEXT: s_add_nc_u64 s[14:15], s[6:7], s[14:15] +; GFX1250-NEXT: s_add_nc_u64 s[14:15], s[4:5], s[14:15] ; GFX1250-NEXT: s_mul_hi_u32 s16, s9, s12 ; GFX1250-NEXT: s_mul_hi_u32 s18, s9, s13 -; GFX1250-NEXT: s_add_co_u32 s6, s14, s17 -; GFX1250-NEXT: s_add_co_ci_u32 s6, s15, s16 +; GFX1250-NEXT: s_add_co_u32 s4, s14, s17 +; GFX1250-NEXT: s_add_co_ci_u32 s4, s15, s16 ; GFX1250-NEXT: s_mul_i32 s12, s9, s13 ; GFX1250-NEXT: s_add_co_ci_u32 s13, s18, 0 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: s_add_nc_u64 s[12:13], s[6:7], s[12:13] +; GFX1250-NEXT: s_add_nc_u64 s[12:13], s[4:5], s[12:13] ; GFX1250-NEXT: s_add_co_u32 s8, s8, s12 -; GFX1250-NEXT: s_cselect_b32 s6, -1, 0 +; GFX1250-NEXT: s_cselect_b32 s4, -1, 0 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1250-NEXT: s_cmp_lg_u32 s4, 0 ; GFX1250-NEXT: s_add_co_ci_u32 s9, s9, s13 ; GFX1250-NEXT: s_mul_u64 s[10:11], s[10:11], s[8:9] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_mul_hi_u32 s13, s8, s11 ; GFX1250-NEXT: s_mul_i32 s12, s8, s11 -; GFX1250-NEXT: s_mul_hi_u32 s6, s8, s10 +; GFX1250-NEXT: s_mul_hi_u32 s4, s8, s10 ; GFX1250-NEXT: s_mul_i32 s15, s9, s10 -; GFX1250-NEXT: s_add_nc_u64 s[12:13], s[6:7], s[12:13] +; GFX1250-NEXT: s_add_nc_u64 s[12:13], s[4:5], s[12:13] ; GFX1250-NEXT: s_mul_hi_u32 s14, s9, s10 ; GFX1250-NEXT: s_mul_hi_u32 s16, s9, s11 -; GFX1250-NEXT: s_add_co_u32 s6, s12, s15 -; GFX1250-NEXT: s_add_co_ci_u32 s6, s13, s14 +; GFX1250-NEXT: s_add_co_u32 s4, s12, s15 +; GFX1250-NEXT: s_add_co_ci_u32 s4, s13, s14 ; GFX1250-NEXT: s_mul_i32 s10, s9, s11 ; GFX1250-NEXT: s_add_co_ci_u32 s11, s16, 0 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: s_add_nc_u64 s[10:11], s[6:7], s[10:11] +; GFX1250-NEXT: s_add_nc_u64 s[10:11], s[4:5], s[10:11] ; GFX1250-NEXT: s_add_co_u32 s8, s8, s10 ; GFX1250-NEXT: s_cselect_b32 s10, -1, 0 -; GFX1250-NEXT: s_mul_hi_u32 s6, s2, s8 +; GFX1250-NEXT: s_mul_hi_u32 s4, s2, s8 ; GFX1250-NEXT: s_cmp_lg_u32 s10, 0 ; GFX1250-NEXT: s_mul_hi_u32 s12, s3, s8 ; GFX1250-NEXT: s_add_co_ci_u32 s10, s9, s11 @@ -3170,33 +3170,33 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) { ; GFX1250-NEXT: s_mul_hi_u32 s9, s2, s10 ; GFX1250-NEXT: s_mul_i32 s8, s2, s10 ; GFX1250-NEXT: s_mul_hi_u32 s13, s3, s10 -; GFX1250-NEXT: s_add_nc_u64 s[8:9], s[6:7], s[8:9] +; GFX1250-NEXT: s_add_nc_u64 s[8:9], s[4:5], s[8:9] ; GFX1250-NEXT: s_mul_i32 s10, s3, s10 -; GFX1250-NEXT: s_add_co_u32 s6, s8, s11 -; GFX1250-NEXT: s_add_co_ci_u32 s6, s9, s12 +; GFX1250-NEXT: s_add_co_u32 s4, s8, s11 +; GFX1250-NEXT: s_add_co_ci_u32 s4, s9, s12 ; GFX1250-NEXT: s_add_co_ci_u32 s11, s13, 0 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: s_add_nc_u64 s[8:9], s[6:7], s[10:11] +; GFX1250-NEXT: s_add_nc_u64 s[8:9], s[4:5], s[10:11] ; GFX1250-NEXT: s_and_b64 s[10:11], s[8:9], 0xffffffff00000000 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: s_or_b32 s10, s10, s8 -; GFX1250-NEXT: s_mul_u64 s[8:9], s[4:5], s[10:11] +; GFX1250-NEXT: s_mul_u64 s[8:9], s[6:7], s[10:11] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-NEXT: s_sub_co_u32 s6, s2, s8 +; GFX1250-NEXT: s_sub_co_u32 s4, s2, s8 ; GFX1250-NEXT: s_cselect_b32 s8, -1, 0 ; GFX1250-NEXT: s_sub_co_i32 s12, s3, s9 ; GFX1250-NEXT: s_cmp_lg_u32 s8, 0 -; GFX1250-NEXT: s_sub_co_ci_u32 s12, s12, s5 -; GFX1250-NEXT: s_sub_co_u32 s13, s6, s4 +; GFX1250-NEXT: s_sub_co_ci_u32 s12, s12, s7 +; GFX1250-NEXT: s_sub_co_u32 s13, s4, s6 ; GFX1250-NEXT: s_cselect_b32 s14, -1, 0 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: s_cmp_lg_u32 s14, 0 ; GFX1250-NEXT: s_sub_co_ci_u32 s12, s12, 0 -; GFX1250-NEXT: s_cmp_ge_u32 s12, s5 +; GFX1250-NEXT: s_cmp_ge_u32 s12, s7 ; GFX1250-NEXT: s_cselect_b32 s14, -1, 0 -; GFX1250-NEXT: s_cmp_ge_u32 s13, s4 +; GFX1250-NEXT: s_cmp_ge_u32 s13, s6 ; GFX1250-NEXT: s_cselect_b32 s15, -1, 0 -; GFX1250-NEXT: s_cmp_eq_u32 s12, s5 +; GFX1250-NEXT: s_cmp_eq_u32 s12, s7 ; GFX1250-NEXT: s_add_nc_u64 s[12:13], s[10:11], 1 ; GFX1250-NEXT: s_cselect_b32 s16, s15, s14 ; GFX1250-NEXT: s_add_nc_u64 s[14:15], s[10:11], 2 @@ -3206,20 +3206,20 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) { ; GFX1250-NEXT: s_cmp_lg_u32 s8, 0 ; GFX1250-NEXT: s_sub_co_ci_u32 s3, s3, s9 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-NEXT: s_cmp_ge_u32 s3, s5 +; GFX1250-NEXT: s_cmp_ge_u32 s3, s7 ; GFX1250-NEXT: s_cselect_b32 s8, -1, 0 -; GFX1250-NEXT: s_cmp_ge_u32 s6, s4 -; GFX1250-NEXT: s_cselect_b32 s6, -1, 0 -; GFX1250-NEXT: s_cmp_eq_u32 s3, s5 -; GFX1250-NEXT: s_cselect_b32 s3, s6, s8 +; GFX1250-NEXT: s_cmp_ge_u32 s4, s6 +; GFX1250-NEXT: s_cselect_b32 s4, -1, 0 +; GFX1250-NEXT: s_cmp_eq_u32 s3, s7 +; GFX1250-NEXT: s_cselect_b32 s3, s4, s8 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: s_cmp_lg_u32 s3, 0 ; GFX1250-NEXT: s_cselect_b32 s9, s13, s11 ; GFX1250-NEXT: s_cselect_b32 s8, s12, s10 ; GFX1250-NEXT: s_cbranch_execnz .LBB16_3 ; GFX1250-NEXT: .LBB16_2: -; GFX1250-NEXT: v_cvt_f32_u32_e32 v0, s4 -; GFX1250-NEXT: s_sub_co_i32 s5, 0, s4 +; GFX1250-NEXT: v_cvt_f32_u32_e32 v0, s6 +; GFX1250-NEXT: s_sub_co_i32 s4, 0, s6 ; GFX1250-NEXT: s_mov_b32 s9, 0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(TRANS32_DEP_1) ; GFX1250-NEXT: v_rcp_iflag_f32_e32 v0, v0 @@ -3228,23 +3228,23 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) { ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_cvt_u32_f32_e32 v0, v0 ; GFX1250-NEXT: v_readfirstlane_b32 s3, v0 -; GFX1250-NEXT: s_mul_i32 s5, s5, s3 +; GFX1250-NEXT: s_mul_i32 s4, s4, s3 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: s_mul_hi_u32 s5, s3, s5 -; GFX1250-NEXT: s_add_co_i32 s3, s3, s5 +; GFX1250-NEXT: s_mul_hi_u32 s4, s3, s4 +; GFX1250-NEXT: s_add_co_i32 s3, s3, s4 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) ; GFX1250-NEXT: s_mul_hi_u32 s3, s2, s3 -; GFX1250-NEXT: s_mul_i32 s5, s3, s4 +; GFX1250-NEXT: s_mul_i32 s4, s3, s6 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-NEXT: s_sub_co_i32 s2, s2, s5 -; GFX1250-NEXT: s_add_co_i32 s5, s3, 1 -; GFX1250-NEXT: s_sub_co_i32 s6, s2, s4 -; GFX1250-NEXT: s_cmp_ge_u32 s2, s4 -; GFX1250-NEXT: s_cselect_b32 s3, s5, s3 -; GFX1250-NEXT: s_cselect_b32 s2, s6, s2 -; GFX1250-NEXT: s_add_co_i32 s5, s3, 1 -; GFX1250-NEXT: s_cmp_ge_u32 s2, s4 -; GFX1250-NEXT: s_cselect_b32 s8, s5, s3 +; GFX1250-NEXT: s_sub_co_i32 s2, s2, s4 +; GFX1250-NEXT: s_add_co_i32 s4, s3, 1 +; GFX1250-NEXT: s_sub_co_i32 s5, s2, s6 +; GFX1250-NEXT: s_cmp_ge_u32 s2, s6 +; GFX1250-NEXT: s_cselect_b32 s3, s4, s3 +; GFX1250-NEXT: s_cselect_b32 s2, s5, s2 +; GFX1250-NEXT: s_add_co_i32 s4, s3, 1 +; GFX1250-NEXT: s_cmp_ge_u32 s2, s6 +; GFX1250-NEXT: s_cselect_b32 s8, s4, s3 ; GFX1250-NEXT: .LBB16_3: ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[8:9] diff --git a/llvm/test/CodeGen/AMDGPU/coalesce-copy-to-agpr-to-av-registers.mir b/llvm/test/CodeGen/AMDGPU/coalesce-copy-to-agpr-to-av-registers.mir index ce1ea4d..c475efb 100644 --- a/llvm/test/CodeGen/AMDGPU/coalesce-copy-to-agpr-to-av-registers.mir +++ b/llvm/test/CodeGen/AMDGPU/coalesce-copy-to-agpr-to-av-registers.mir @@ -504,7 +504,7 @@ body: | ; CHECK-NEXT: [[COPY1:%[0-9]+]].sub1:areg_64 = COPY [[COPY]].sub0 ; CHECK-NEXT: [[COPY:%[0-9]+]].sub1:vreg_64 = COPY [[COPY]].sub0 ; CHECK-NEXT: INLINEASM &"; use $0", 0 /* attdialect */, 4325385 /* reguse:AReg_64 */, [[COPY1]] - ; CHECK-NEXT: INLINEASM &"; use $0", 0 /* attdialect */, 3670025 /* reguse:VS_64_with_sub1 */, [[COPY]] + ; CHECK-NEXT: INLINEASM &"; use $0", 0 /* attdialect */, 3735561 /* reguse:VReg_64 */, [[COPY]] ; CHECK-NEXT: SI_RETURN %0:vgpr_32 = COPY $vgpr0 undef %1.sub0:areg_64 = COPY %0 @@ -512,7 +512,7 @@ body: | undef %2.sub0:vreg_64 = COPY %0 %2.sub1:vreg_64 = COPY %0 INLINEASM &"; use $0", 0 /* attdialect */, 4325385 /* reguse:AReg_64 */, killed %1 - INLINEASM &"; use $0", 0 /* attdialect */, 3670025 /* reguse:VReg_64 */, killed %2 + INLINEASM &"; use $0", 0 /* attdialect */, 3735561 /* reguse:VReg_64 */, killed %2 SI_RETURN ... diff --git a/llvm/test/CodeGen/AMDGPU/ds_write2.ll b/llvm/test/CodeGen/AMDGPU/ds_write2.ll index be60a00..0cae0e5 100644 --- a/llvm/test/CodeGen/AMDGPU/ds_write2.ll +++ b/llvm/test/CodeGen/AMDGPU/ds_write2.ll @@ -705,12 +705,13 @@ define amdgpu_kernel void @write2_ptr_subreg_arg_two_val_f32(ptr addrspace(1) %C ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b32 v1, v0, s[0:1] scale_offset -; GFX1250-NEXT: global_load_b32 v0, v0, s[2:3] scale_offset -; GFX1250-NEXT: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s5 +; GFX1250-NEXT: global_load_b32 v2, v0, s[2:3] scale_offset +; GFX1250-NEXT: s_wait_xcnt 0x0 +; GFX1250-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v3, s5 ; GFX1250-NEXT: s_wait_loadcnt 0x1 -; GFX1250-NEXT: ds_store_b32 v2, v1 offset:32 +; GFX1250-NEXT: ds_store_b32 v0, v1 offset:32 ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: ds_store_b32 v3, v0 offset:32 +; GFX1250-NEXT: ds_store_b32 v3, v2 offset:32 ; GFX1250-NEXT: s_endpgm %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1 %in0.gep = getelementptr float, ptr addrspace(1) %in0, i32 %x.i @@ -1282,14 +1283,14 @@ define amdgpu_kernel void @simple_write2_v4f32_superreg_align4(ptr addrspace(3) ; ; GFX1250-LABEL: simple_write2_v4f32_superreg_align4: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b64 s[0:1], s[4:5], 0x8 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b32 s4, s[4:5], 0x0 +; GFX1250-NEXT: s_clause 0x1 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x8 +; GFX1250-NEXT: s_load_b32 s8, s[4:5], 0x0 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_load_b128 s[0:3], s[0:1], 0x0 +; GFX1250-NEXT: s_load_b128 s[0:3], s[6:7], 0x0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-NEXT: v_lshl_add_u32 v0, v0, 4, s4 +; GFX1250-NEXT: v_lshl_add_u32 v0, v0, 4, s8 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_dual_mov_b32 v1, s2 :: v_dual_mov_b32 v2, s3 ; GFX1250-NEXT: v_dual_mov_b32 v3, s0 :: v_dual_mov_b32 v4, s1 diff --git a/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll b/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll index 1e7855c..eefc781 100644 --- a/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll +++ b/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll @@ -541,11 +541,10 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-LABEL: flat_xchg_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -570,9 +569,8 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB10_2 ; GFX1250-SDAG-NEXT: .LBB10_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: s_clause 0x1 @@ -586,14 +584,13 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; ; GFX1250-GISEL-LABEL: flat_xchg_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -618,10 +615,9 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB10_2 ; GFX1250-GISEL-NEXT: .LBB10_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: s_clause 0x1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off @@ -727,13 +723,12 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB11_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -754,9 +749,8 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB11_2 ; GFX1250-SDAG-NEXT: .LBB11_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: s_clause 0x1 @@ -770,8 +764,7 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; ; GFX1250-GISEL-LABEL: flat_xchg_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -780,7 +773,7 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -805,10 +798,9 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB11_2 ; GFX1250-GISEL-NEXT: .LBB11_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: s_clause 0x1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off @@ -917,11 +909,10 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -943,9 +934,8 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB12_2 ; GFX1250-SDAG-NEXT: .LBB12_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v0, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_store_b64 v0, v[2:3], off @@ -953,15 +943,14 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_xchg_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB12_3 @@ -982,10 +971,9 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB12_2 ; GFX1250-GISEL-NEXT: .LBB12_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v0, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_store_b64 v0, v[4:5], off ; GFX1250-GISEL-NEXT: s_endpgm @@ -1069,11 +1057,9 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB13_3 @@ -1094,9 +1080,8 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB13_2 ; GFX1250-SDAG-NEXT: .LBB13_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v0, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_store_b64 v0, v[2:3], off @@ -1104,8 +1089,7 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; ; GFX1250-GISEL-LABEL: flat_xchg_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -1115,7 +1099,7 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB13_3 @@ -1136,10 +1120,9 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB13_2 ; GFX1250-GISEL-NEXT: .LBB13_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v0, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_store_b64 v0, v[4:5], off ; GFX1250-GISEL-NEXT: s_endpgm @@ -1400,11 +1383,10 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_add_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -1429,9 +1411,8 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB18_2 ; GFX1250-SDAG-NEXT: .LBB18_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -1445,14 +1426,13 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_add_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -1477,10 +1457,9 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB18_2 ; GFX1250-GISEL-NEXT: .LBB18_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -1590,13 +1569,12 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB19_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -1617,9 +1595,8 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB19_2 ; GFX1250-SDAG-NEXT: .LBB19_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -1633,8 +1610,7 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_add_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -1643,7 +1619,7 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -1668,10 +1644,9 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB19_2 ; GFX1250-GISEL-NEXT: .LBB19_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -1784,11 +1759,10 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -1810,9 +1784,8 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB20_2 ; GFX1250-SDAG-NEXT: .LBB20_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -1823,15 +1796,14 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_add_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB20_3 @@ -1852,10 +1824,9 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB20_2 ; GFX1250-GISEL-NEXT: .LBB20_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -1950,11 +1921,9 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB21_3 @@ -1975,9 +1944,8 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB21_2 ; GFX1250-SDAG-NEXT: .LBB21_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -1988,8 +1956,7 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_add_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -1999,7 +1966,7 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB21_3 @@ -2020,10 +1987,9 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB21_2 ; GFX1250-GISEL-NEXT: .LBB21_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -2295,11 +2261,10 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_sub_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -2324,9 +2289,8 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB26_2 ; GFX1250-SDAG-NEXT: .LBB26_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -2340,14 +2304,13 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_sub_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -2372,10 +2335,9 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB26_2 ; GFX1250-GISEL-NEXT: .LBB26_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -2487,13 +2449,12 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB27_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -2514,9 +2475,8 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB27_2 ; GFX1250-SDAG-NEXT: .LBB27_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -2530,8 +2490,7 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_sub_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -2540,7 +2499,7 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -2565,10 +2524,9 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB27_2 ; GFX1250-GISEL-NEXT: .LBB27_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -2683,11 +2641,10 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -2709,9 +2666,8 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB28_2 ; GFX1250-SDAG-NEXT: .LBB28_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -2722,15 +2678,14 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_sub_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB28_3 @@ -2751,10 +2706,9 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB28_2 ; GFX1250-GISEL-NEXT: .LBB28_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -2851,11 +2805,9 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB29_3 @@ -2876,9 +2828,8 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB29_2 ; GFX1250-SDAG-NEXT: .LBB29_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -2889,8 +2840,7 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_sub_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -2900,7 +2850,7 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB29_3 @@ -2921,10 +2871,9 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB29_2 ; GFX1250-GISEL-NEXT: .LBB29_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -3198,11 +3147,10 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_and_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -3227,9 +3175,8 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB34_2 ; GFX1250-SDAG-NEXT: .LBB34_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -3244,14 +3191,13 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_and_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -3276,10 +3222,9 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB34_2 ; GFX1250-GISEL-NEXT: .LBB34_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -3390,13 +3335,12 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB35_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -3417,9 +3361,8 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB35_2 ; GFX1250-SDAG-NEXT: .LBB35_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -3434,8 +3377,7 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_and_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -3444,7 +3386,7 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -3469,10 +3411,9 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB35_2 ; GFX1250-GISEL-NEXT: .LBB35_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -3586,11 +3527,10 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -3612,9 +3552,8 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB36_2 ; GFX1250-SDAG-NEXT: .LBB36_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -3626,15 +3565,14 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_and_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB36_3 @@ -3655,10 +3593,9 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB36_2 ; GFX1250-GISEL-NEXT: .LBB36_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -3754,11 +3691,9 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB37_3 @@ -3779,9 +3714,8 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB37_2 ; GFX1250-SDAG-NEXT: .LBB37_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -3793,8 +3727,7 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_and_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -3804,7 +3737,7 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB37_3 @@ -3825,10 +3758,9 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB37_2 ; GFX1250-GISEL-NEXT: .LBB37_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -4101,11 +4033,10 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn(ptr inreg %sbase, i32 %voffs ; GFX1250-SDAG-LABEL: flat_or_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -4130,9 +4061,8 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn(ptr inreg %sbase, i32 %voffs ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB42_2 ; GFX1250-SDAG-NEXT: .LBB42_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -4147,14 +4077,13 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn(ptr inreg %sbase, i32 %voffs ; ; GFX1250-GISEL-LABEL: flat_or_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -4179,10 +4108,9 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn(ptr inreg %sbase, i32 %voffs ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB42_2 ; GFX1250-GISEL-NEXT: .LBB42_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -4293,13 +4221,12 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn_neg128(ptr inreg %sbase, i32 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB43_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -4320,9 +4247,8 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn_neg128(ptr inreg %sbase, i32 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB43_2 ; GFX1250-SDAG-NEXT: .LBB43_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -4337,8 +4263,7 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn_neg128(ptr inreg %sbase, i32 ; ; GFX1250-GISEL-LABEL: flat_or_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -4347,7 +4272,7 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn_neg128(ptr inreg %sbase, i32 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -4372,10 +4297,9 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn_neg128(ptr inreg %sbase, i32 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB43_2 ; GFX1250-GISEL-NEXT: .LBB43_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -4489,11 +4413,10 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, i ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -4515,9 +4438,8 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, i ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB44_2 ; GFX1250-SDAG-NEXT: .LBB44_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -4529,15 +4451,14 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, i ; ; GFX1250-GISEL-LABEL: flat_or_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB44_3 @@ -4558,10 +4479,9 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, i ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB44_2 ; GFX1250-GISEL-NEXT: .LBB44_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -4657,11 +4577,9 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB45_3 @@ -4682,9 +4600,8 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB45_2 ; GFX1250-SDAG-NEXT: .LBB45_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -4696,8 +4613,7 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vof ; ; GFX1250-GISEL-LABEL: flat_or_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -4707,7 +4623,7 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vof ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB45_3 @@ -4728,10 +4644,9 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vof ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB45_2 ; GFX1250-GISEL-NEXT: .LBB45_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -5004,11 +4919,10 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_xor_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -5033,9 +4947,8 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB50_2 ; GFX1250-SDAG-NEXT: .LBB50_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -5050,14 +4963,13 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_xor_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -5082,10 +4994,9 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB50_2 ; GFX1250-GISEL-NEXT: .LBB50_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -5196,13 +5107,12 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB51_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -5223,9 +5133,8 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB51_2 ; GFX1250-SDAG-NEXT: .LBB51_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -5240,8 +5149,7 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_xor_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -5250,7 +5158,7 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -5275,10 +5183,9 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB51_2 ; GFX1250-GISEL-NEXT: .LBB51_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -5392,11 +5299,10 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -5418,9 +5324,8 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB52_2 ; GFX1250-SDAG-NEXT: .LBB52_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -5432,15 +5337,14 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_xor_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB52_3 @@ -5461,10 +5365,9 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB52_2 ; GFX1250-GISEL-NEXT: .LBB52_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -5560,11 +5463,9 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB53_3 @@ -5585,9 +5486,8 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB53_2 ; GFX1250-SDAG-NEXT: .LBB53_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -5599,8 +5499,7 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_xor_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -5610,7 +5509,7 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB53_3 @@ -5631,10 +5530,9 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB53_2 ; GFX1250-GISEL-NEXT: .LBB53_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -5877,11 +5775,10 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_max_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -5902,9 +5799,8 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB58_2 ; GFX1250-SDAG-NEXT: .LBB58_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -5918,14 +5814,13 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_max_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -5946,10 +5841,9 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB58_2 ; GFX1250-GISEL-NEXT: .LBB58_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -6061,13 +5955,12 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB59_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -6084,9 +5977,8 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB59_2 ; GFX1250-SDAG-NEXT: .LBB59_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -6100,8 +5992,7 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_max_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -6110,7 +6001,7 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -6131,10 +6022,9 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB59_2 ; GFX1250-GISEL-NEXT: .LBB59_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -6249,11 +6139,10 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -6272,9 +6161,8 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB60_2 ; GFX1250-SDAG-NEXT: .LBB60_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -6285,15 +6173,14 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_max_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB60_3 @@ -6311,10 +6198,9 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB60_2 ; GFX1250-GISEL-NEXT: .LBB60_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -6409,11 +6295,9 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB61_3 @@ -6431,9 +6315,8 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB61_2 ; GFX1250-SDAG-NEXT: .LBB61_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -6444,8 +6327,7 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_max_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -6455,7 +6337,7 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB61_3 @@ -6473,10 +6355,9 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB61_2 ; GFX1250-GISEL-NEXT: .LBB61_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -6718,11 +6599,10 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_min_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -6743,9 +6623,8 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB66_2 ; GFX1250-SDAG-NEXT: .LBB66_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -6759,14 +6638,13 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_min_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -6787,10 +6665,9 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB66_2 ; GFX1250-GISEL-NEXT: .LBB66_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -6902,13 +6779,12 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB67_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -6925,9 +6801,8 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB67_2 ; GFX1250-SDAG-NEXT: .LBB67_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -6941,8 +6816,7 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_min_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -6951,7 +6825,7 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -6972,10 +6846,9 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB67_2 ; GFX1250-GISEL-NEXT: .LBB67_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -7090,11 +6963,10 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -7113,9 +6985,8 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB68_2 ; GFX1250-SDAG-NEXT: .LBB68_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -7126,15 +6997,14 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_min_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB68_3 @@ -7152,10 +7022,9 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB68_2 ; GFX1250-GISEL-NEXT: .LBB68_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -7250,11 +7119,9 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB69_3 @@ -7272,9 +7139,8 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB69_2 ; GFX1250-SDAG-NEXT: .LBB69_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -7285,8 +7151,7 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_min_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -7296,7 +7161,7 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB69_3 @@ -7314,10 +7179,9 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB69_2 ; GFX1250-GISEL-NEXT: .LBB69_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -7559,11 +7423,10 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-LABEL: flat_umax_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -7584,9 +7447,8 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB74_2 ; GFX1250-SDAG-NEXT: .LBB74_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -7600,14 +7462,13 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; ; GFX1250-GISEL-LABEL: flat_umax_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -7628,10 +7489,9 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB74_2 ; GFX1250-GISEL-NEXT: .LBB74_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -7743,13 +7603,12 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB75_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -7766,9 +7625,8 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB75_2 ; GFX1250-SDAG-NEXT: .LBB75_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -7782,8 +7640,7 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; ; GFX1250-GISEL-LABEL: flat_umax_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -7792,7 +7649,7 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -7813,10 +7670,9 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB75_2 ; GFX1250-GISEL-NEXT: .LBB75_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -7931,11 +7787,10 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -7954,9 +7809,8 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB76_2 ; GFX1250-SDAG-NEXT: .LBB76_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -7967,15 +7821,14 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_umax_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB76_3 @@ -7993,10 +7846,9 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB76_2 ; GFX1250-GISEL-NEXT: .LBB76_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -8091,11 +7943,9 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB77_3 @@ -8113,9 +7963,8 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB77_2 ; GFX1250-SDAG-NEXT: .LBB77_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -8126,8 +7975,7 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; ; GFX1250-GISEL-LABEL: flat_umax_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -8137,7 +7985,7 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB77_3 @@ -8155,10 +8003,9 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB77_2 ; GFX1250-GISEL-NEXT: .LBB77_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -8400,11 +8247,10 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-LABEL: flat_umin_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -8425,9 +8271,8 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB82_2 ; GFX1250-SDAG-NEXT: .LBB82_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -8441,14 +8286,13 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; ; GFX1250-GISEL-LABEL: flat_umin_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -8469,10 +8313,9 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn(ptr inreg %sbase, i32 %vof ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB82_2 ; GFX1250-GISEL-NEXT: .LBB82_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -8584,13 +8427,12 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB83_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -8607,9 +8449,8 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB83_2 ; GFX1250-SDAG-NEXT: .LBB83_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -8623,8 +8464,7 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; ; GFX1250-GISEL-LABEL: flat_umin_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -8633,7 +8473,7 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -8654,10 +8494,9 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn_neg128(ptr inreg %sbase, i ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB83_2 ; GFX1250-GISEL-NEXT: .LBB83_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -8772,11 +8611,10 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -8795,9 +8633,8 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB84_2 ; GFX1250-SDAG-NEXT: .LBB84_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -8808,15 +8645,14 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_umin_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB84_3 @@ -8834,10 +8670,9 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB84_2 ; GFX1250-GISEL-NEXT: .LBB84_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -8932,11 +8767,9 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB85_3 @@ -8954,9 +8787,8 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB85_2 ; GFX1250-SDAG-NEXT: .LBB85_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -8967,8 +8799,7 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; ; GFX1250-GISEL-LABEL: flat_umin_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -8978,7 +8809,7 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB85_3 @@ -8996,10 +8827,9 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB85_2 ; GFX1250-GISEL-NEXT: .LBB85_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -9281,12 +9111,11 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn(ptr inreg %sbase, i32 % ; GFX1250-SDAG-LABEL: flat_cmpxchg_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v7, v2 :: v_dual_mov_b32 v6, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v5, v4 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v4, v3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[2:3], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v3 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -9311,9 +9140,8 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn(ptr inreg %sbase, i32 % ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB90_2 ; GFX1250-SDAG-NEXT: .LBB90_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v2 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v8, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v8, off @@ -9328,15 +9156,14 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn(ptr inreg %sbase, i32 % ; ; GFX1250-GISEL-LABEL: flat_cmpxchg_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v0 :: v_dual_mov_b32 v10, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v8, v1 :: v_dual_mov_b32 v9, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v0 :: v_dual_mov_b32 v8, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, v3 :: v_dual_mov_b32 v7, v4 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v9, v2 :: v_dual_mov_b32 v6, v3 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v0, v5 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v3, v10 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v7, v4 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v3 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -9361,10 +9188,9 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn(ptr inreg %sbase, i32 % ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB90_2 ; GFX1250-GISEL-NEXT: .LBB90_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v4, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -9485,13 +9311,12 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn_neg128(ptr inreg %sbase ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[2:3], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v3 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v3 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB91_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -9512,9 +9337,8 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn_neg128(ptr inreg %sbase ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB91_2 ; GFX1250-SDAG-NEXT: .LBB91_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v2 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v8, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v8, off @@ -9529,10 +9353,9 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn_neg128(ptr inreg %sbase ; ; GFX1250-GISEL-LABEL: flat_cmpxchg_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v0 :: v_dual_mov_b32 v10, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v8, v1 :: v_dual_mov_b32 v9, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v0 :: v_dual_mov_b32 v8, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, v3 :: v_dual_mov_b32 v7, v4 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v9, v2 :: v_dual_mov_b32 v6, v3 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v5 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v1, null, 0, v1, vcc_lo @@ -9540,7 +9363,7 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn_neg128(ptr inreg %sbase ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v3, v10 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v7, v4 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v3 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -9565,10 +9388,9 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn_neg128(ptr inreg %sbase ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB91_2 ; GFX1250-GISEL-NEXT: .LBB91_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v4, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -9690,13 +9512,12 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffs ; GFX1250-SDAG-LABEL: flat_cmpxchg_saddr_i64_nortn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v7, v2 :: v_dual_mov_b32 v6, v1 -; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v5, v4 :: v_dual_mov_b32 v4, v3 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v5, v4 +; GFX1250-SDAG-NEXT: v_mov_b32_e32 v4, v3 +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v2, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v2, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v2 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -9718,9 +9539,8 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffs ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB92_2 ; GFX1250-SDAG-NEXT: .LBB92_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v2, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v2, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v2, -1, v2, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v2, off @@ -9732,15 +9552,15 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffs ; ; GFX1250-GISEL-LABEL: flat_cmpxchg_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v9, v2 :: v_dual_mov_b32 v6, v3 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v8, v1 :: v_dual_mov_b32 v9, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, v3 :: v_dual_mov_b32 v7, v4 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, src_flat_scratch_base_hi :: v_dual_mov_b32 v8, v1 ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v7, v4 :: v_dual_bitop2_b32 v1, v3, v5 bitop3:0x14 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB92_3 @@ -9761,10 +9581,9 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffs ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB92_2 ; GFX1250-GISEL-NEXT: .LBB92_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -9869,11 +9688,9 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v2, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v2, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v2 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB93_3 @@ -9894,9 +9711,8 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB93_2 ; GFX1250-SDAG-NEXT: .LBB93_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v2, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v2, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v2, -1, v2, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v2, off @@ -9908,18 +9724,18 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 ; ; GFX1250-GISEL-LABEL: flat_cmpxchg_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v9, v2 :: v_dual_mov_b32 v6, v3 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v8, v1 :: v_dual_mov_b32 v9, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, v3 :: v_dual_mov_b32 v7, v4 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, src_flat_scratch_base_hi :: v_dual_mov_b32 v8, v1 ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v1, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v7, v4 :: v_dual_bitop2_b32 v1, v3, v5 bitop3:0x14 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB93_3 @@ -9940,10 +9756,9 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB93_2 ; GFX1250-GISEL-NEXT: .LBB93_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -10188,11 +10003,10 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_inc_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -10214,10 +10028,9 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB98_2 ; GFX1250-SDAG-NEXT: .LBB98_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v6, off @@ -10233,14 +10046,13 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_inc_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -10262,18 +10074,17 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB98_2 ; GFX1250-GISEL-NEXT: .LBB98_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 ; GFX1250-GISEL-NEXT: v_add_nc_u64_e32 v[2:3], 1, v[0:1] ; GFX1250-GISEL-NEXT: v_cmp_ge_u64_e32 vcc_lo, v[0:1], v[4:5] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v2, v2, 0, vcc_lo +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v3, v3, 0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_store_b64 v6, v[2:3], off ; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 @@ -10386,13 +10197,12 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB99_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -10410,10 +10220,9 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB99_2 ; GFX1250-SDAG-NEXT: .LBB99_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v6, off @@ -10429,8 +10238,7 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_inc_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -10439,7 +10247,7 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -10461,18 +10269,17 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB99_2 ; GFX1250-GISEL-NEXT: .LBB99_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 ; GFX1250-GISEL-NEXT: v_add_nc_u64_e32 v[2:3], 1, v[0:1] ; GFX1250-GISEL-NEXT: v_cmp_ge_u64_e32 vcc_lo, v[0:1], v[4:5] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v2, v2, 0, vcc_lo +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v3, v3, 0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_store_b64 v6, v[2:3], off ; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 @@ -10588,11 +10395,10 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -10610,9 +10416,8 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB100_2 ; GFX1250-SDAG-NEXT: .LBB100_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v6, off @@ -10625,15 +10430,14 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_inc_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB100_3 @@ -10650,17 +10454,16 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB100_2 ; GFX1250-GISEL-NEXT: .LBB100_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 ; GFX1250-GISEL-NEXT: v_add_nc_u64_e32 v[2:3], 1, v[0:1] ; GFX1250-GISEL-NEXT: v_cmp_ge_u64_e32 vcc_lo, v[0:1], v[4:5] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v0, v2, 0, vcc_lo +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v1, v3, 0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_store_b64 v6, v[0:1], off ; GFX1250-GISEL-NEXT: s_endpgm @@ -10754,11 +10557,9 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB101_3 @@ -10775,9 +10576,8 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB101_2 ; GFX1250-SDAG-NEXT: .LBB101_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v6, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v6, off @@ -10790,8 +10590,7 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_inc_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -10801,7 +10600,7 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB101_3 @@ -10818,17 +10617,16 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB101_2 ; GFX1250-GISEL-NEXT: .LBB101_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_2) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 ; GFX1250-GISEL-NEXT: v_add_nc_u64_e32 v[2:3], 1, v[0:1] ; GFX1250-GISEL-NEXT: v_cmp_ge_u64_e32 vcc_lo, v[0:1], v[4:5] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v0, v2, 0, vcc_lo +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_3) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v1, v3, 0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_store_b64 v6, v[0:1], off ; GFX1250-GISEL-NEXT: s_endpgm @@ -11064,11 +10862,10 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-LABEL: flat_dec_saddr_i64_rtn: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -11090,10 +10887,9 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s1, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB106_2 ; GFX1250-SDAG-NEXT: .LBB106_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s0, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v6, off @@ -11112,14 +10908,13 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; ; GFX1250-GISEL-LABEL: flat_dec_saddr_i64_rtn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, v0, v3 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, 0, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -11141,11 +10936,10 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn(ptr inreg %sbase, i32 %voff ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s1, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB106_2 ; GFX1250-GISEL-NEXT: .LBB106_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -11268,13 +11062,12 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[4:5], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v5 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GFX1250-SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-SDAG-NEXT: s_and_saveexec_b32 s0, vcc_lo +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB107_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %Flow @@ -11292,10 +11085,9 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s1, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB107_2 ; GFX1250-SDAG-NEXT: .LBB107_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, s0, v4 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v6, off @@ -11314,8 +11106,7 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; ; GFX1250-GISEL-LABEL: flat_dec_saddr_i64_rtn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v8, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v3, v0 :: v_dual_mov_b32 v4, v1 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, v0, v3 @@ -11324,7 +11115,7 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: v_add_co_u32 v6, vcc_lo, 0xffffff80, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v7, null, -1, v1, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v7, v8 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v5, v2 :: v_dual_bitop2_b32 v0, src_flat_scratch_base_hi, v7 bitop3:0x14 ; GFX1250-GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX1250-GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo @@ -11346,11 +11137,10 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn_neg128(ptr inreg %sbase, i3 ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s1, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB107_2 ; GFX1250-GISEL-NEXT: .LBB107_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v6, v0 +; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v6 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v6, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v6, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -11476,11 +11266,10 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v2, v1 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v1, 0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 @@ -11498,9 +11287,8 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB108_2 ; GFX1250-SDAG-NEXT: .LBB108_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -11516,15 +11304,14 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; ; GFX1250-GISEL-LABEL: flat_dec_saddr_i64_nortn: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, v2, v0 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, 0, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB108_3 @@ -11541,10 +11328,9 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB108_2 ; GFX1250-GISEL-NEXT: .LBB108_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -11648,11 +11434,9 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[2:3], v[0:1] ; GFX1250-SDAG-NEXT: v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, s0, v1 ; GFX1250-SDAG-NEXT: s_mov_b32 s0, exec_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-SDAG-NEXT: v_xor_b32_e32 v4, src_flat_scratch_base_hi, v1 ; GFX1250-SDAG-NEXT: v_cmpx_lt_u32_e32 0x3ffffff, v4 ; GFX1250-SDAG-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB109_3 @@ -11669,9 +11453,8 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-SDAG-NEXT: s_cbranch_execz .LBB109_2 ; GFX1250-SDAG-NEXT: .LBB109_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1] -; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, s0, v0 +; GFX1250-SDAG-NEXT: v_subrev_nc_u32_e32 v4, src_flat_scratch_base_lo, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v4, vcc_lo ; GFX1250-SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -11687,8 +11470,7 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; ; GFX1250-GISEL-LABEL: flat_dec_saddr_i64_nortn_neg128: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v6, src_flat_scratch_base_hi :: v_dual_mov_b32 v4, v1 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v5, v2 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v4, v1 :: v_dual_mov_b32 v5, v2 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] ; GFX1250-GISEL-NEXT: s_mov_b32 s0, exec_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) @@ -11698,7 +11480,7 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: v_add_co_u32 v2, vcc_lo, 0xffffff80, v1 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v3, null, -1, v3, vcc_lo ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, v3, v6 +; GFX1250-GISEL-NEXT: v_xor_b32_e32 v1, src_flat_scratch_base_hi, v3 ; GFX1250-GISEL-NEXT: v_cmpx_le_u32_e32 0x4000000, v1 ; GFX1250-GISEL-NEXT: s_xor_b32 s0, exec_lo, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execnz .LBB109_3 @@ -11715,10 +11497,9 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo ; GFX1250-GISEL-NEXT: s_and_not1_saveexec_b32 s0, s0 ; GFX1250-GISEL-NEXT: s_cbranch_execz .LBB109_2 ; GFX1250-GISEL-NEXT: .LBB109_4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_sub_nc_u32_e32 v0, v2, v0 +; GFX1250-GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_cndmask_b32_e32 v2, -1, v0, vcc_lo ; GFX1250-GISEL-NEXT: scratch_load_b64 v[0:1], v2, off ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -11834,12 +11615,10 @@ define double @flat_atomic_fadd_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_and_not1_b32 vcc_lo, exec_lo, s2 ; GFX1250-SDAG-NEXT: s_cbranch_vccz .LBB110_3 ; GFX1250-SDAG-NEXT: ; %bb.1: ; %atomicrmw.check.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_cmp_lt_u32 s2, 0x4000000 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, -1, 0 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_and_not1_b32 vcc_lo, exec_lo, s2 ; GFX1250-SDAG-NEXT: s_cbranch_vccz .LBB110_4 ; GFX1250-SDAG-NEXT: ; %bb.2: ; %atomicrmw.global @@ -11855,9 +11634,7 @@ define double @flat_atomic_fadd_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: .LBB110_4: ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr2_vgpr3 ; GFX1250-SDAG-NEXT: .LBB110_5: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, s2 +; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, s2, -1 ; GFX1250-SDAG-NEXT: scratch_load_b64 v[2:3], off, s2 @@ -11891,10 +11668,9 @@ define double @flat_atomic_fadd_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr2_vgpr3 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB110_6 ; GFX1250-GISEL-NEXT: ; %bb.1: ; %atomicrmw.check.private -; GFX1250-GISEL-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi +; GFX1250-GISEL-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr2_vgpr3 -; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-GISEL-NEXT: s_cmp_ge_u32 s2, 0x4000000 ; GFX1250-GISEL-NEXT: s_mov_b32 s2, 1 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB110_3 @@ -11910,9 +11686,7 @@ define double @flat_atomic_fadd_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_cmp_lg_u32 s2, 0 ; GFX1250-GISEL-NEXT: s_cbranch_scc1 .LBB110_5 ; GFX1250-GISEL-NEXT: ; %bb.4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_sub_co_i32 s2, s0, s2 +; GFX1250-GISEL-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-GISEL-NEXT: s_cselect_b32 s2, s2, -1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[2:3], off, s2 @@ -12060,12 +11834,10 @@ define void @flat_atomic_fadd_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: .LBB111_2: ; %atomicrmw.phi ; GFX1250-SDAG-NEXT: s_set_pc_i64 s[30:31] ; GFX1250-SDAG-NEXT: .LBB111_3: ; %atomicrmw.check.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi +; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_cmp_lt_u32 s2, 0x4000000 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, -1, 0 -; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_and_not1_b32 vcc_lo, exec_lo, s2 ; GFX1250-SDAG-NEXT: s_mov_b32 s2, -1 ; GFX1250-SDAG-NEXT: s_cbranch_vccz .LBB111_5 @@ -12079,9 +11851,7 @@ define void @flat_atomic_fadd_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_and_not1_b32 vcc_lo, exec_lo, s2 ; GFX1250-SDAG-NEXT: s_cbranch_vccnz .LBB111_7 ; GFX1250-SDAG-NEXT: ; %bb.6: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, s2 +; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, s2, -1 ; GFX1250-SDAG-NEXT: scratch_load_b64 v[2:3], off, s2 @@ -12112,9 +11882,8 @@ define void @flat_atomic_fadd_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_cmp_lg_u32 s1, s3 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB111_6 ; GFX1250-GISEL-NEXT: ; %bb.1: ; %atomicrmw.check.private -; GFX1250-GISEL-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-GISEL-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi +; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-GISEL-NEXT: s_cmp_ge_u32 s2, 0x4000000 ; GFX1250-GISEL-NEXT: s_mov_b32 s2, 1 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB111_3 @@ -12130,9 +11899,7 @@ define void @flat_atomic_fadd_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_cmp_lg_u32 s2, 0 ; GFX1250-GISEL-NEXT: s_cbranch_scc1 .LBB111_5 ; GFX1250-GISEL-NEXT: ; %bb.4: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_sub_co_i32 s2, s0, s2 +; GFX1250-GISEL-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-GISEL-NEXT: s_cselect_b32 s2, s2, -1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[2:3], off, s2 @@ -12261,9 +12028,8 @@ define double @flat_atomic_fmax_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_add_nc_u64 s[0:1], s[0:1], 0x50 -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_cmp_lt_u32 s2, 0x4000000 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, -1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) @@ -12279,10 +12045,9 @@ define double @flat_atomic_fmax_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: .LBB112_2: ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr2_vgpr3 ; GFX1250-SDAG-NEXT: .LBB112_3: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-SDAG-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, s2 +; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 +; GFX1250-SDAG-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-SDAG-NEXT: s_cselect_b32 s0, s2, -1 ; GFX1250-SDAG-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x0 @@ -12301,10 +12066,9 @@ define double @flat_atomic_fmax_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_add_co_u32 s2, s0, 0x50 ; GFX1250-GISEL-NEXT: s_add_co_ci_u32 s3, s1, 0 -; GFX1250-GISEL-NEXT: s_mov_b32 s4, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr2_vgpr3 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, s4 +; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: s_cmp_ge_u32 s4, 0x4000000 ; GFX1250-GISEL-NEXT: s_mov_b32 s4, 1 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB112_2 @@ -12320,10 +12084,9 @@ define double @flat_atomic_fmax_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_cmp_lg_u32 s0, 0 ; GFX1250-GISEL-NEXT: s_cbranch_scc1 .LBB112_4 ; GFX1250-GISEL-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo -; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, s0 +; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-GISEL-NEXT: s_cselect_b32 s0, s0, -1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -12413,9 +12176,8 @@ define void @flat_atomic_fmax_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_add_nc_u64 s[0:1], s[0:1], 0x50 -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_cmp_lt_u32 s2, 0x4000000 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, -1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) @@ -12434,11 +12196,10 @@ define void @flat_atomic_fmax_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_wait_storecnt_dscnt 0x0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB113_2 ; GFX1250-SDAG-NEXT: .LBB113_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo +; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo +; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 ; GFX1250-SDAG-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, s2 -; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-SDAG-NEXT: s_cselect_b32 s0, s2, -1 ; GFX1250-SDAG-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x0 @@ -12454,9 +12215,8 @@ define void @flat_atomic_fmax_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_add_co_u32 s2, s0, 0x50 ; GFX1250-GISEL-NEXT: s_add_co_ci_u32 s3, s1, 0 -; GFX1250-GISEL-NEXT: s_mov_b32 s4, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, s4 +; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: s_cmp_ge_u32 s4, 0x4000000 ; GFX1250-GISEL-NEXT: s_mov_b32 s4, 1 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB113_2 @@ -12473,10 +12233,9 @@ define void @flat_atomic_fmax_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_cmp_lg_u32 s0, 0 ; GFX1250-GISEL-NEXT: s_cbranch_scc1 .LBB113_4 ; GFX1250-GISEL-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo -; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, s0 +; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-GISEL-NEXT: s_cselect_b32 s0, s0, -1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -12562,9 +12321,8 @@ define double @flat_atomic_fmin_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_add_nc_u64 s[0:1], s[0:1], 0x50 -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_cmp_lt_u32 s2, 0x4000000 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, -1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) @@ -12580,10 +12338,9 @@ define double @flat_atomic_fmin_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: .LBB114_2: ; GFX1250-SDAG-NEXT: ; implicit-def: $vgpr2_vgpr3 ; GFX1250-SDAG-NEXT: .LBB114_3: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo -; GFX1250-SDAG-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, s2 +; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo ; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 +; GFX1250-SDAG-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-SDAG-NEXT: s_cselect_b32 s0, s2, -1 ; GFX1250-SDAG-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x0 @@ -12602,10 +12359,9 @@ define double @flat_atomic_fmin_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_add_co_u32 s2, s0, 0x50 ; GFX1250-GISEL-NEXT: s_add_co_ci_u32 s3, s1, 0 -; GFX1250-GISEL-NEXT: s_mov_b32 s4, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: ; implicit-def: $vgpr2_vgpr3 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, s4 +; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: s_cmp_ge_u32 s4, 0x4000000 ; GFX1250-GISEL-NEXT: s_mov_b32 s4, 1 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB114_2 @@ -12621,10 +12377,9 @@ define double @flat_atomic_fmin_f64_saddr_rtn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_cmp_lg_u32 s0, 0 ; GFX1250-GISEL-NEXT: s_cbranch_scc1 .LBB114_4 ; GFX1250-GISEL-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo -; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, s0 +; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-GISEL-NEXT: s_cselect_b32 s0, s0, -1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 @@ -12714,9 +12469,8 @@ define void @flat_atomic_fmin_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_add_nc_u64 s[0:1], s[0:1], 0x50 -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, s2 +; GFX1250-SDAG-NEXT: s_xor_b32 s2, s1, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_cmp_lt_u32 s2, 0x4000000 ; GFX1250-SDAG-NEXT: s_cselect_b32 s2, -1, 0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) @@ -12735,11 +12489,10 @@ define void @flat_atomic_fmin_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-SDAG-NEXT: s_wait_storecnt_dscnt 0x0 ; GFX1250-SDAG-NEXT: s_cbranch_execnz .LBB115_2 ; GFX1250-SDAG-NEXT: .LBB115_4: ; %atomicrmw.private -; GFX1250-SDAG-NEXT: s_mov_b32 s2, src_flat_scratch_base_lo +; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, src_flat_scratch_base_lo +; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 ; GFX1250-SDAG-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-SDAG-NEXT: s_sub_co_i32 s2, s0, s2 -; GFX1250-SDAG-NEXT: s_cmp_lg_u64 s[0:1], 0 ; GFX1250-SDAG-NEXT: s_cselect_b32 s0, s2, -1 ; GFX1250-SDAG-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x0 @@ -12755,9 +12508,8 @@ define void @flat_atomic_fmin_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_add_co_u32 s2, s0, 0x50 ; GFX1250-GISEL-NEXT: s_add_co_ci_u32 s3, s1, 0 -; GFX1250-GISEL-NEXT: s_mov_b32 s4, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, s4 +; GFX1250-GISEL-NEXT: s_xor_b32 s4, s3, src_flat_scratch_base_hi ; GFX1250-GISEL-NEXT: s_cmp_ge_u32 s4, 0x4000000 ; GFX1250-GISEL-NEXT: s_mov_b32 s4, 1 ; GFX1250-GISEL-NEXT: s_cbranch_scc0 .LBB115_2 @@ -12774,10 +12526,9 @@ define void @flat_atomic_fmin_f64_saddr_nortn(ptr inreg %ptr, double %data) { ; GFX1250-GISEL-NEXT: s_cmp_lg_u32 s0, 0 ; GFX1250-GISEL-NEXT: s_cbranch_scc1 .LBB115_4 ; GFX1250-GISEL-NEXT: ; %bb.3: ; %atomicrmw.private -; GFX1250-GISEL-NEXT: s_mov_b32 s0, src_flat_scratch_base_lo -; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] -; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, s0 +; GFX1250-GISEL-NEXT: s_sub_co_i32 s0, s2, src_flat_scratch_base_lo ; GFX1250-GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1250-GISEL-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-GISEL-NEXT: s_cselect_b32 s0, s0, -1 ; GFX1250-GISEL-NEXT: scratch_load_b64 v[2:3], off, s0 ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 diff --git a/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll b/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll index 2079543..b5b2655 100644 --- a/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll +++ b/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll @@ -2179,6 +2179,7 @@ define amdgpu_ps void @flat_addr_64bit_lsr_iv_multiload(ptr inreg %arg, ptr inre ; GFX1250-SDAG-NEXT: flat_load_b32 v1, v0, s[4:5] scope:SCOPE_SYS ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x0 ; GFX1250-SDAG-NEXT: s_cmp_eq_u32 s0, 0x400 +; GFX1250-SDAG-NEXT: ; kill: killed $sgpr4_sgpr5 ; GFX1250-SDAG-NEXT: s_cbranch_scc0 .LBB117_1 ; GFX1250-SDAG-NEXT: ; %bb.2: ; %bb2 ; GFX1250-SDAG-NEXT: s_endpgm @@ -2190,15 +2191,16 @@ define amdgpu_ps void @flat_addr_64bit_lsr_iv_multiload(ptr inreg %arg, ptr inre ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[0:1] ; GFX1250-GISEL-NEXT: .LBB117_1: ; %bb3 ; GFX1250-GISEL-NEXT: ; =>This Inner Loop Header: Depth=1 -; GFX1250-GISEL-NEXT: s_wait_dscnt 0x0 ; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_add_co_u32 v4, vcc_lo, v0, v2 ; GFX1250-GISEL-NEXT: v_add_co_ci_u32_e64 v5, null, v1, v3, vcc_lo ; GFX1250-GISEL-NEXT: v_add_nc_u64_e32 v[2:3], 4, v[2:3] +; GFX1250-GISEL-NEXT: ; kill: killed $vgpr4 killed $vgpr5 +; GFX1250-GISEL-NEXT: s_wait_dscnt 0x0 +; GFX1250-GISEL-NEXT: flat_load_b32 v6, v[4:5] scope:SCOPE_SYS +; GFX1250-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 ; GFX1250-GISEL-NEXT: flat_load_b32 v6, v[4:5] scope:SCOPE_SYS -; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 -; GFX1250-GISEL-NEXT: flat_load_b32 v4, v[4:5] scope:SCOPE_SYS ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 ; GFX1250-GISEL-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0x400, v2 ; GFX1250-GISEL-NEXT: s_cbranch_vccz .LBB117_1 diff --git a/llvm/test/CodeGen/AMDGPU/fmax3.ll b/llvm/test/CodeGen/AMDGPU/fmax3.ll index 4827f75..5e6de6d 100644 --- a/llvm/test/CodeGen/AMDGPU/fmax3.ll +++ b/llvm/test/CodeGen/AMDGPU/fmax3.ll @@ -162,32 +162,32 @@ define amdgpu_kernel void @test_fmax3_olt_0_f32(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-LABEL: test_fmax3_olt_0_f32: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s18, s10 -; GFX1250-NEXT: s_mov_b32 s19, s11 -; GFX1250-NEXT: s_mov_b32 s22, s10 -; GFX1250-NEXT: s_mov_b32 s23, s11 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s6, s2 +; GFX1250-NEXT: s_mov_b32 s7, s3 +; GFX1250-NEXT: s_mov_b32 s18, s2 +; GFX1250-NEXT: s_mov_b32 s19, s3 +; GFX1250-NEXT: s_mov_b32 s22, s2 +; GFX1250-NEXT: s_mov_b32 s23, s3 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s12, s2 -; GFX1250-NEXT: s_mov_b32 s13, s3 -; GFX1250-NEXT: s_mov_b32 s16, s4 -; GFX1250-NEXT: s_mov_b32 s17, s5 -; GFX1250-NEXT: s_mov_b32 s20, s6 -; GFX1250-NEXT: s_mov_b32 s21, s7 -; GFX1250-NEXT: buffer_load_b32 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s10 +; GFX1250-NEXT: s_mov_b32 s5, s11 +; GFX1250-NEXT: s_mov_b32 s16, s12 +; GFX1250-NEXT: s_mov_b32 s17, s13 +; GFX1250-NEXT: s_mov_b32 s20, s14 +; GFX1250-NEXT: s_mov_b32 s21, s15 +; GFX1250-NEXT: buffer_load_b32 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: s_mov_b32 s0, s8 +; GFX1250-NEXT: s_mov_b32 s1, s9 ; GFX1250-NEXT: v_max3_num_f32 v0, v0, v1, v2 -; GFX1250-NEXT: buffer_store_b32 v0, off, s[8:11], null +; GFX1250-NEXT: buffer_store_b32 v0, off, s[0:3], null ; GFX1250-NEXT: s_endpgm %a = load volatile float, ptr addrspace(1) %aptr, align 4 %b = load volatile float, ptr addrspace(1) %bptr, align 4 @@ -352,32 +352,32 @@ define amdgpu_kernel void @test_fmax3_olt_1_f32(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-LABEL: test_fmax3_olt_1_f32: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s18, s10 -; GFX1250-NEXT: s_mov_b32 s19, s11 -; GFX1250-NEXT: s_mov_b32 s22, s10 -; GFX1250-NEXT: s_mov_b32 s23, s11 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s6, s2 +; GFX1250-NEXT: s_mov_b32 s7, s3 +; GFX1250-NEXT: s_mov_b32 s18, s2 +; GFX1250-NEXT: s_mov_b32 s19, s3 +; GFX1250-NEXT: s_mov_b32 s22, s2 +; GFX1250-NEXT: s_mov_b32 s23, s3 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s12, s2 -; GFX1250-NEXT: s_mov_b32 s13, s3 -; GFX1250-NEXT: s_mov_b32 s16, s4 -; GFX1250-NEXT: s_mov_b32 s17, s5 -; GFX1250-NEXT: s_mov_b32 s20, s6 -; GFX1250-NEXT: s_mov_b32 s21, s7 -; GFX1250-NEXT: buffer_load_b32 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s10 +; GFX1250-NEXT: s_mov_b32 s5, s11 +; GFX1250-NEXT: s_mov_b32 s16, s12 +; GFX1250-NEXT: s_mov_b32 s17, s13 +; GFX1250-NEXT: s_mov_b32 s20, s14 +; GFX1250-NEXT: s_mov_b32 s21, s15 +; GFX1250-NEXT: buffer_load_b32 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: s_mov_b32 s0, s8 +; GFX1250-NEXT: s_mov_b32 s1, s9 ; GFX1250-NEXT: v_max3_num_f32 v0, v2, v0, v1 -; GFX1250-NEXT: buffer_store_b32 v0, off, s[8:11], null +; GFX1250-NEXT: buffer_store_b32 v0, off, s[0:3], null ; GFX1250-NEXT: s_endpgm %a = load volatile float, ptr addrspace(1) %aptr, align 4 %b = load volatile float, ptr addrspace(1) %bptr, align 4 @@ -609,62 +609,62 @@ define amdgpu_kernel void @test_fmax3_olt_0_f16(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-TRUE16-LABEL: test_fmax3_olt_0_f16: ; GFX1250-TRUE16: ; %bb.0: -; GFX1250-TRUE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-TRUE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-TRUE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-TRUE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-TRUE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-TRUE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-TRUE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-TRUE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-TRUE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-TRUE16-NEXT: buffer_load_u16 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-TRUE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-TRUE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-TRUE16-NEXT: buffer_load_u16 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-TRUE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-TRUE16-NEXT: v_max3_num_f16 v0.l, v0.l, v1.l, v2.l -; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-TRUE16-NEXT: s_endpgm ; ; GFX1250-FAKE16-LABEL: test_fmax3_olt_0_f16: ; GFX1250-FAKE16: ; %bb.0: -; GFX1250-FAKE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-FAKE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-FAKE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-FAKE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-FAKE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-FAKE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-FAKE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-FAKE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-FAKE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-FAKE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-FAKE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-FAKE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-FAKE16-NEXT: v_max3_num_f16 v0, v0, v1, v2 -; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-FAKE16-NEXT: s_endpgm %a = load volatile half, ptr addrspace(1) %aptr, align 2 %b = load volatile half, ptr addrspace(1) %bptr, align 2 @@ -897,62 +897,62 @@ define amdgpu_kernel void @test_fmax3_olt_1_f16(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-TRUE16-LABEL: test_fmax3_olt_1_f16: ; GFX1250-TRUE16: ; %bb.0: -; GFX1250-TRUE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-TRUE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-TRUE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-TRUE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-TRUE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-TRUE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-TRUE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-TRUE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-TRUE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-TRUE16-NEXT: buffer_load_u16 v1, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-TRUE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-TRUE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-TRUE16-NEXT: buffer_load_u16 v1, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v2, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v0, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-TRUE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-TRUE16-NEXT: v_max3_num_f16 v0.l, v0.l, v1.l, v2.l -; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-TRUE16-NEXT: s_endpgm ; ; GFX1250-FAKE16-LABEL: test_fmax3_olt_1_f16: ; GFX1250-FAKE16: ; %bb.0: -; GFX1250-FAKE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-FAKE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-FAKE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-FAKE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-FAKE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-FAKE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-FAKE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-FAKE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-FAKE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-FAKE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-FAKE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-FAKE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-FAKE16-NEXT: v_max3_num_f16 v0, v2, v0, v1 -; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-FAKE16-NEXT: s_endpgm %a = load volatile half, ptr addrspace(1) %aptr, align 2 %b = load volatile half, ptr addrspace(1) %bptr, align 2 diff --git a/llvm/test/CodeGen/AMDGPU/fmin3.ll b/llvm/test/CodeGen/AMDGPU/fmin3.ll index 6dfefd8..6a6f232 100644 --- a/llvm/test/CodeGen/AMDGPU/fmin3.ll +++ b/llvm/test/CodeGen/AMDGPU/fmin3.ll @@ -162,32 +162,32 @@ define amdgpu_kernel void @test_fmin3_olt_0_f32(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-LABEL: test_fmin3_olt_0_f32: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s18, s10 -; GFX1250-NEXT: s_mov_b32 s19, s11 -; GFX1250-NEXT: s_mov_b32 s22, s10 -; GFX1250-NEXT: s_mov_b32 s23, s11 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s6, s2 +; GFX1250-NEXT: s_mov_b32 s7, s3 +; GFX1250-NEXT: s_mov_b32 s18, s2 +; GFX1250-NEXT: s_mov_b32 s19, s3 +; GFX1250-NEXT: s_mov_b32 s22, s2 +; GFX1250-NEXT: s_mov_b32 s23, s3 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s12, s2 -; GFX1250-NEXT: s_mov_b32 s13, s3 -; GFX1250-NEXT: s_mov_b32 s16, s4 -; GFX1250-NEXT: s_mov_b32 s17, s5 -; GFX1250-NEXT: s_mov_b32 s20, s6 -; GFX1250-NEXT: s_mov_b32 s21, s7 -; GFX1250-NEXT: buffer_load_b32 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s10 +; GFX1250-NEXT: s_mov_b32 s5, s11 +; GFX1250-NEXT: s_mov_b32 s16, s12 +; GFX1250-NEXT: s_mov_b32 s17, s13 +; GFX1250-NEXT: s_mov_b32 s20, s14 +; GFX1250-NEXT: s_mov_b32 s21, s15 +; GFX1250-NEXT: buffer_load_b32 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: s_mov_b32 s0, s8 +; GFX1250-NEXT: s_mov_b32 s1, s9 ; GFX1250-NEXT: v_min3_num_f32 v0, v0, v1, v2 -; GFX1250-NEXT: buffer_store_b32 v0, off, s[8:11], null +; GFX1250-NEXT: buffer_store_b32 v0, off, s[0:3], null ; GFX1250-NEXT: s_endpgm %a = load volatile float, ptr addrspace(1) %aptr, align 4 %b = load volatile float, ptr addrspace(1) %bptr, align 4 @@ -352,32 +352,32 @@ define amdgpu_kernel void @test_fmin3_olt_1_f32(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-LABEL: test_fmin3_olt_1_f32: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s18, s10 -; GFX1250-NEXT: s_mov_b32 s19, s11 -; GFX1250-NEXT: s_mov_b32 s22, s10 -; GFX1250-NEXT: s_mov_b32 s23, s11 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s6, s2 +; GFX1250-NEXT: s_mov_b32 s7, s3 +; GFX1250-NEXT: s_mov_b32 s18, s2 +; GFX1250-NEXT: s_mov_b32 s19, s3 +; GFX1250-NEXT: s_mov_b32 s22, s2 +; GFX1250-NEXT: s_mov_b32 s23, s3 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s12, s2 -; GFX1250-NEXT: s_mov_b32 s13, s3 -; GFX1250-NEXT: s_mov_b32 s16, s4 -; GFX1250-NEXT: s_mov_b32 s17, s5 -; GFX1250-NEXT: s_mov_b32 s20, s6 -; GFX1250-NEXT: s_mov_b32 s21, s7 -; GFX1250-NEXT: buffer_load_b32 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s10 +; GFX1250-NEXT: s_mov_b32 s5, s11 +; GFX1250-NEXT: s_mov_b32 s16, s12 +; GFX1250-NEXT: s_mov_b32 s17, s13 +; GFX1250-NEXT: s_mov_b32 s20, s14 +; GFX1250-NEXT: s_mov_b32 s21, s15 +; GFX1250-NEXT: buffer_load_b32 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b32 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: s_mov_b32 s0, s8 +; GFX1250-NEXT: s_mov_b32 s1, s9 ; GFX1250-NEXT: v_min3_num_f32 v0, v2, v0, v1 -; GFX1250-NEXT: buffer_store_b32 v0, off, s[8:11], null +; GFX1250-NEXT: buffer_store_b32 v0, off, s[0:3], null ; GFX1250-NEXT: s_endpgm %a = load volatile float, ptr addrspace(1) %aptr, align 4 %b = load volatile float, ptr addrspace(1) %bptr, align 4 @@ -609,62 +609,62 @@ define amdgpu_kernel void @test_fmin3_olt_0_f16(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-TRUE16-LABEL: test_fmin3_olt_0_f16: ; GFX1250-TRUE16: ; %bb.0: -; GFX1250-TRUE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-TRUE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-TRUE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-TRUE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-TRUE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-TRUE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-TRUE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-TRUE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-TRUE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-TRUE16-NEXT: buffer_load_u16 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-TRUE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-TRUE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-TRUE16-NEXT: buffer_load_u16 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-TRUE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-TRUE16-NEXT: v_min3_num_f16 v0.l, v0.l, v1.l, v2.l -; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-TRUE16-NEXT: s_endpgm ; ; GFX1250-FAKE16-LABEL: test_fmin3_olt_0_f16: ; GFX1250-FAKE16: ; %bb.0: -; GFX1250-FAKE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-FAKE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-FAKE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-FAKE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-FAKE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-FAKE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-FAKE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-FAKE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-FAKE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-FAKE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-FAKE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-FAKE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-FAKE16-NEXT: v_min3_num_f16 v0, v0, v1, v2 -; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-FAKE16-NEXT: s_endpgm %a = load volatile half, ptr addrspace(1) %aptr, align 2 %b = load volatile half, ptr addrspace(1) %bptr, align 2 @@ -897,62 +897,62 @@ define amdgpu_kernel void @test_fmin3_olt_1_f16(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-TRUE16-LABEL: test_fmin3_olt_1_f16: ; GFX1250-TRUE16: ; %bb.0: -; GFX1250-TRUE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-TRUE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-TRUE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-TRUE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-TRUE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-TRUE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-TRUE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-TRUE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-TRUE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-TRUE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-TRUE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-TRUE16-NEXT: buffer_load_u16 v1, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-TRUE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-TRUE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-TRUE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-TRUE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-TRUE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-TRUE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-TRUE16-NEXT: buffer_load_u16 v1, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v2, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-TRUE16-NEXT: buffer_load_u16 v0, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-TRUE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-TRUE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-TRUE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-TRUE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-TRUE16-NEXT: v_min3_num_f16 v0.l, v0.l, v1.l, v2.l -; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-TRUE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-TRUE16-NEXT: s_endpgm ; ; GFX1250-FAKE16-LABEL: test_fmin3_olt_1_f16: ; GFX1250-FAKE16: ; %bb.0: -; GFX1250-FAKE16-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-FAKE16-NEXT: s_mov_b32 s10, -1 -; GFX1250-FAKE16-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-FAKE16-NEXT: s_mov_b32 s14, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s15, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s11 -; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s10 -; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s11 +; GFX1250-FAKE16-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-FAKE16-NEXT: s_mov_b32 s2, -1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-FAKE16-NEXT: s_mov_b32 s6, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s7, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s18, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s19, s3 +; GFX1250-FAKE16-NEXT: s_mov_b32 s22, s2 +; GFX1250-FAKE16-NEXT: s_mov_b32 s23, s3 ; GFX1250-FAKE16-NEXT: s_wait_kmcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s12, s2 -; GFX1250-FAKE16-NEXT: s_mov_b32 s13, s3 -; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s4 -; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s5 -; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s6 -; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s7 -; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[12:15], null scope:SCOPE_SYS +; GFX1250-FAKE16-NEXT: s_mov_b32 s4, s10 +; GFX1250-FAKE16-NEXT: s_mov_b32 s5, s11 +; GFX1250-FAKE16-NEXT: s_mov_b32 s16, s12 +; GFX1250-FAKE16-NEXT: s_mov_b32 s17, s13 +; GFX1250-FAKE16-NEXT: s_mov_b32 s20, s14 +; GFX1250-FAKE16-NEXT: s_mov_b32 s21, s15 +; GFX1250-FAKE16-NEXT: buffer_load_u16 v0, off, s[4:7], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v1, off, s[16:19], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 ; GFX1250-FAKE16-NEXT: buffer_load_u16 v2, off, s[20:23], null scope:SCOPE_SYS ; GFX1250-FAKE16-NEXT: s_wait_loadcnt 0x0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s8, s0 -; GFX1250-FAKE16-NEXT: s_mov_b32 s9, s1 +; GFX1250-FAKE16-NEXT: s_mov_b32 s0, s8 +; GFX1250-FAKE16-NEXT: s_mov_b32 s1, s9 ; GFX1250-FAKE16-NEXT: v_min3_num_f16 v0, v2, v0, v1 -; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[8:11], null +; GFX1250-FAKE16-NEXT: buffer_store_b16 v0, off, s[0:3], null ; GFX1250-FAKE16-NEXT: s_endpgm %a = load volatile half, ptr addrspace(1) %aptr, align 2 %b = load volatile half, ptr addrspace(1) %bptr, align 2 @@ -1217,36 +1217,36 @@ define amdgpu_kernel void @test_fmin3_olt_0_f64(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-LABEL: test_fmin3_olt_0_f64: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s18, s10 -; GFX1250-NEXT: s_mov_b32 s19, s11 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s6, s2 +; GFX1250-NEXT: s_mov_b32 s7, s3 +; GFX1250-NEXT: s_mov_b32 s18, s2 +; GFX1250-NEXT: s_mov_b32 s19, s3 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s12, s2 -; GFX1250-NEXT: s_mov_b32 s13, s3 -; GFX1250-NEXT: s_mov_b32 s16, s4 -; GFX1250-NEXT: s_mov_b32 s17, s5 -; GFX1250-NEXT: buffer_load_b64 v[0:1], off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s10 +; GFX1250-NEXT: s_mov_b32 s5, s11 +; GFX1250-NEXT: s_mov_b32 s16, s12 +; GFX1250-NEXT: s_mov_b32 s17, s13 +; GFX1250-NEXT: buffer_load_b64 v[0:1], off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b64 v[2:3], off, s[16:19], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: s_wait_xcnt 0x1 -; GFX1250-NEXT: s_mov_b32 s12, s6 -; GFX1250-NEXT: s_mov_b32 s13, s7 -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: buffer_load_b64 v[4:5], off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s14 +; GFX1250-NEXT: s_mov_b32 s5, s15 +; GFX1250-NEXT: s_mov_b32 s0, s8 +; GFX1250-NEXT: buffer_load_b64 v[4:5], off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: s_mov_b32 s1, s9 ; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-NEXT: v_max_num_f64_e32 v[2:3], v[2:3], v[2:3] ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_min_num_f64_e32 v[0:1], v[0:1], v[2:3] ; GFX1250-NEXT: v_max_num_f64_e32 v[2:3], v[4:5], v[4:5] ; GFX1250-NEXT: v_min_num_f64_e32 v[0:1], v[0:1], v[2:3] -; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[8:11], null +; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[0:3], null ; GFX1250-NEXT: s_endpgm %a = load volatile double, ptr addrspace(1) %aptr, align 4 %b = load volatile double, ptr addrspace(1) %bptr, align 4 @@ -1427,36 +1427,36 @@ define amdgpu_kernel void @test_fmin3_olt_1_f64(ptr addrspace(1) %out, ptr addrs ; ; GFX1250-LABEL: test_fmin3_olt_1_f64: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s18, s10 -; GFX1250-NEXT: s_mov_b32 s19, s11 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 +; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s6, s2 +; GFX1250-NEXT: s_mov_b32 s7, s3 +; GFX1250-NEXT: s_mov_b32 s18, s2 +; GFX1250-NEXT: s_mov_b32 s19, s3 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s12, s2 -; GFX1250-NEXT: s_mov_b32 s13, s3 -; GFX1250-NEXT: s_mov_b32 s16, s4 -; GFX1250-NEXT: s_mov_b32 s17, s5 -; GFX1250-NEXT: buffer_load_b64 v[0:1], off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s10 +; GFX1250-NEXT: s_mov_b32 s5, s11 +; GFX1250-NEXT: s_mov_b32 s16, s12 +; GFX1250-NEXT: s_mov_b32 s17, s13 +; GFX1250-NEXT: buffer_load_b64 v[0:1], off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: buffer_load_b64 v[2:3], off, s[16:19], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: s_wait_xcnt 0x1 -; GFX1250-NEXT: s_mov_b32 s12, s6 -; GFX1250-NEXT: s_mov_b32 s13, s7 -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: buffer_load_b64 v[4:5], off, s[12:15], null scope:SCOPE_SYS +; GFX1250-NEXT: s_mov_b32 s4, s14 +; GFX1250-NEXT: s_mov_b32 s5, s15 +; GFX1250-NEXT: s_mov_b32 s0, s8 +; GFX1250-NEXT: buffer_load_b64 v[4:5], off, s[4:7], null scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: s_mov_b32 s1, s9 ; GFX1250-NEXT: v_max_num_f64_e32 v[0:1], v[0:1], v[0:1] ; GFX1250-NEXT: v_max_num_f64_e32 v[2:3], v[2:3], v[2:3] ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_min_num_f64_e32 v[0:1], v[0:1], v[2:3] ; GFX1250-NEXT: v_max_num_f64_e32 v[2:3], v[4:5], v[4:5] ; GFX1250-NEXT: v_min_num_f64_e32 v[0:1], v[2:3], v[0:1] -; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[8:11], null +; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[0:3], null ; GFX1250-NEXT: s_endpgm %a = load volatile double, ptr addrspace(1) %aptr, align 4 %b = load volatile double, ptr addrspace(1) %bptr, align 4 diff --git a/llvm/test/CodeGen/AMDGPU/global-load-xcnt.ll b/llvm/test/CodeGen/AMDGPU/global-load-xcnt.ll index e532dea..f807169 100644 --- a/llvm/test/CodeGen/AMDGPU/global-load-xcnt.ll +++ b/llvm/test/CodeGen/AMDGPU/global-load-xcnt.ll @@ -11,22 +11,20 @@ define void @test_i8load_v4i8store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-SDAG: ; %bb.0: ; GCN-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 ; GCN-SDAG-NEXT: s_wait_kmcnt 0x0 -; GCN-SDAG-NEXT: global_load_u8 v2, v[2:3], off -; GCN-SDAG-NEXT: global_load_u8 v3, v[4:5], off -; GCN-SDAG-NEXT: global_load_u8 v0, v[0:1], off +; GCN-SDAG-NEXT: global_load_u8 v6, v[2:3], off +; GCN-SDAG-NEXT: global_load_u8 v7, v[4:5], off +; GCN-SDAG-NEXT: global_load_u8 v10, v[0:1], off ; GCN-SDAG-NEXT: s_wait_loadcnt 0x2 ; GCN-SDAG-NEXT: s_wait_xcnt 0x0 -; GCN-SDAG-NEXT: v_lshlrev_b16 v1, 8, v2 +; GCN-SDAG-NEXT: v_lshlrev_b16 v0, 8, v6 ; GCN-SDAG-NEXT: s_wait_loadcnt 0x1 -; GCN-SDAG-NEXT: v_lshlrev_b16 v2, 8, v3 +; GCN-SDAG-NEXT: v_lshlrev_b16 v1, 8, v7 +; GCN-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) +; GCN-SDAG-NEXT: v_or_b32_e32 v1, v7, v1 ; GCN-SDAG-NEXT: s_wait_loadcnt 0x0 -; GCN-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) -; GCN-SDAG-NEXT: v_or_b32_e32 v0, v0, v1 -; GCN-SDAG-NEXT: v_or_b32_e32 v1, v3, v2 -; GCN-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_3) -; GCN-SDAG-NEXT: v_lshlrev_b32_e32 v1, 16, v1 +; GCN-SDAG-NEXT: v_dual_lshlrev_b32 v1, 16, v1 :: v_dual_bitop2_b32 v0, v10, v0 bitop3:0x54 +; GCN-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GCN-SDAG-NEXT: v_and_b32_e32 v0, 0xffff, v0 -; GCN-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GCN-SDAG-NEXT: v_or_b32_e32 v0, v0, v1 ; GCN-SDAG-NEXT: global_store_b32 v[8:9], v0, off ; GCN-SDAG-NEXT: s_set_pc_i64 s[30:31] @@ -35,13 +33,15 @@ define void @test_i8load_v4i8store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-GISEL: ; %bb.0: ; GCN-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 ; GCN-GISEL-NEXT: s_wait_kmcnt 0x0 -; GCN-GISEL-NEXT: global_load_u8 v0, v[0:1], off -; GCN-GISEL-NEXT: global_load_u8 v1, v[2:3], off -; GCN-GISEL-NEXT: global_load_u8 v2, v[4:5], off +; GCN-GISEL-NEXT: global_load_u8 v6, v[0:1], off +; GCN-GISEL-NEXT: global_load_u8 v7, v[2:3], off +; GCN-GISEL-NEXT: global_load_u8 v10, v[4:5], off ; GCN-GISEL-NEXT: s_wait_loadcnt 0x1 -; GCN-GISEL-NEXT: v_lshl_or_b32 v0, v1, 8, v0 +; GCN-GISEL-NEXT: s_wait_xcnt 0x2 +; GCN-GISEL-NEXT: v_lshl_or_b32 v0, v7, 8, v6 ; GCN-GISEL-NEXT: s_wait_loadcnt 0x0 -; GCN-GISEL-NEXT: v_dual_lshlrev_b32 v1, 16, v2 :: v_dual_lshlrev_b32 v2, 24, v2 +; GCN-GISEL-NEXT: s_wait_xcnt 0x1 +; GCN-GISEL-NEXT: v_dual_lshlrev_b32 v1, 16, v10 :: v_dual_lshlrev_b32 v2, 24, v10 ; GCN-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GCN-GISEL-NEXT: v_or3_b32 v0, v0, v1, v2 ; GCN-GISEL-NEXT: global_store_b32 v[8:9], v0, off @@ -64,21 +64,21 @@ define i16 @test_v7i16_load_store(ptr addrspace(1) %ptr1, ptr addrspace(1) %ptr2 ; GCN-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 ; GCN-SDAG-NEXT: s_wait_kmcnt 0x0 ; GCN-SDAG-NEXT: global_load_b128 v[4:7], v[0:1], off -; GCN-SDAG-NEXT: global_load_b128 v[0:3], v[2:3], off -; GCN-SDAG-NEXT: v_mov_b64_e32 v[8:9], 0 -; GCN-SDAG-NEXT: s_wait_loadcnt 0x0 -; GCN-SDAG-NEXT: v_pk_add_u16 v10, v6, v2 -; GCN-SDAG-NEXT: v_pk_add_u16 v11, v7, v3 +; GCN-SDAG-NEXT: global_load_b128 v[8:11], v[2:3], off ; GCN-SDAG-NEXT: s_wait_xcnt 0x0 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[2:3], 12 +; GCN-SDAG-NEXT: s_wait_loadcnt 0x0 +; GCN-SDAG-NEXT: v_pk_add_u16 v1, v6, v10 +; GCN-SDAG-NEXT: v_pk_add_u16 v12, v7, v11 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[6:7], 8 -; GCN-SDAG-NEXT: v_pk_add_u16 v4, v4, v0 -; GCN-SDAG-NEXT: v_lshrrev_b32_e32 v0, 16, v10 -; GCN-SDAG-NEXT: v_pk_add_u16 v5, v5, v1 +; GCN-SDAG-NEXT: v_mov_b64_e32 v[10:11], 0 +; GCN-SDAG-NEXT: v_pk_add_u16 v5, v5, v9 +; GCN-SDAG-NEXT: v_lshrrev_b32_e32 v0, 16, v1 +; GCN-SDAG-NEXT: v_pk_add_u16 v4, v4, v8 ; GCN-SDAG-NEXT: s_clause 0x2 -; GCN-SDAG-NEXT: global_store_b16 v[2:3], v11, off -; GCN-SDAG-NEXT: global_store_b32 v[6:7], v10, off -; GCN-SDAG-NEXT: global_store_b64 v[8:9], v[4:5], off +; GCN-SDAG-NEXT: global_store_b16 v[2:3], v12, off +; GCN-SDAG-NEXT: global_store_b32 v[6:7], v1, off +; GCN-SDAG-NEXT: global_store_b64 v[10:11], v[4:5], off ; GCN-SDAG-NEXT: s_set_pc_i64 s[30:31] ; ; GCN-GISEL-LABEL: test_v7i16_load_store: @@ -86,28 +86,29 @@ define i16 @test_v7i16_load_store(ptr addrspace(1) %ptr1, ptr addrspace(1) %ptr2 ; GCN-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 ; GCN-GISEL-NEXT: s_wait_kmcnt 0x0 ; GCN-GISEL-NEXT: global_load_b128 v[4:7], v[0:1], off -; GCN-GISEL-NEXT: global_load_b128 v[0:3], v[2:3], off -; GCN-GISEL-NEXT: v_mov_b64_e32 v[8:9], 0 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[10:11], 2 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[12:13], 4 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[14:15], 6 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[16:17], 8 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[18:19], 10 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[20:21], 12 +; GCN-GISEL-NEXT: global_load_b128 v[8:11], v[2:3], off +; GCN-GISEL-NEXT: s_wait_xcnt 0x0 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[2:3], 0 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[12:13], 2 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[14:15], 4 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[16:17], 6 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[18:19], 8 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[20:21], 10 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[22:23], 12 ; GCN-GISEL-NEXT: s_wait_loadcnt 0x0 -; GCN-GISEL-NEXT: v_pk_add_u16 v2, v6, v2 -; GCN-GISEL-NEXT: v_pk_add_u16 v4, v4, v0 -; GCN-GISEL-NEXT: v_pk_add_u16 v1, v5, v1 -; GCN-GISEL-NEXT: v_pk_add_u16 v3, v7, v3 +; GCN-GISEL-NEXT: v_pk_add_u16 v1, v6, v10 +; GCN-GISEL-NEXT: v_pk_add_u16 v4, v4, v8 +; GCN-GISEL-NEXT: v_pk_add_u16 v5, v5, v9 +; GCN-GISEL-NEXT: v_pk_add_u16 v6, v7, v11 ; GCN-GISEL-NEXT: s_clause 0x6 -; GCN-GISEL-NEXT: global_store_b16 v[8:9], v4, off -; GCN-GISEL-NEXT: global_store_d16_hi_b16 v[10:11], v4, off -; GCN-GISEL-NEXT: global_store_b16 v[12:13], v1, off -; GCN-GISEL-NEXT: global_store_d16_hi_b16 v[14:15], v1, off -; GCN-GISEL-NEXT: global_store_b16 v[16:17], v2, off -; GCN-GISEL-NEXT: global_store_d16_hi_b16 v[18:19], v2, off -; GCN-GISEL-NEXT: global_store_b16 v[20:21], v3, off -; GCN-GISEL-NEXT: v_lshrrev_b32_e32 v0, 16, v2 +; GCN-GISEL-NEXT: global_store_b16 v[2:3], v4, off +; GCN-GISEL-NEXT: global_store_d16_hi_b16 v[12:13], v4, off +; GCN-GISEL-NEXT: global_store_b16 v[14:15], v5, off +; GCN-GISEL-NEXT: global_store_d16_hi_b16 v[16:17], v5, off +; GCN-GISEL-NEXT: global_store_b16 v[18:19], v1, off +; GCN-GISEL-NEXT: global_store_d16_hi_b16 v[20:21], v1, off +; GCN-GISEL-NEXT: global_store_b16 v[22:23], v6, off +; GCN-GISEL-NEXT: v_lshrrev_b32_e32 v0, 16, v1 ; GCN-GISEL-NEXT: s_set_pc_i64 s[30:31] %vec1 = load <7 x i16>, ptr addrspace(1) %ptr1 %insert = insertelement <7 x i16> %vec1, i16 20, i32 4 @@ -253,8 +254,8 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-SDAG-NEXT: global_load_b128 v[22:25], v[0:1], off offset:32 ; GCN-SDAG-NEXT: global_load_b128 v[26:29], v[0:1], off offset:16 ; GCN-SDAG-NEXT: global_load_b128 v[30:33], v[0:1], off -; GCN-SDAG-NEXT: global_load_b128 v[0:3], v[0:1], off offset:64 -; GCN-SDAG-NEXT: v_mov_b64_e32 v[36:37], 0x70 +; GCN-SDAG-NEXT: global_load_b128 v[34:37], v[0:1], off offset:64 +; GCN-SDAG-NEXT: v_mov_b64_e32 v[2:3], 0x70 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[48:49], 48 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[38:39], 0x60 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[50:51], 32 @@ -262,14 +263,15 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-SDAG-NEXT: v_mov_b64_e32 v[66:67], 0 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[52:53], 0x50 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[54:55], 64 -; GCN-SDAG-NEXT: v_dual_mov_b32 v34, 0xc8 :: v_dual_mov_b32 v35, 0 +; GCN-SDAG-NEXT: s_wait_xcnt 0x0 +; GCN-SDAG-NEXT: v_dual_mov_b32 v0, 0xc8 :: v_dual_mov_b32 v1, 0 ; GCN-SDAG-NEXT: s_wait_loadcnt 0x7 -; GCN-SDAG-NEXT: global_store_b128 v[36:37], v[6:9], off +; GCN-SDAG-NEXT: global_store_b128 v[2:3], v[6:9], off ; GCN-SDAG-NEXT: s_wait_loadcnt 0x6 ; GCN-SDAG-NEXT: global_store_b128 v[38:39], v[10:13], off ; GCN-SDAG-NEXT: s_wait_loadcnt 0x5 ; GCN-SDAG-NEXT: s_wait_xcnt 0x1 -; GCN-SDAG-NEXT: v_dual_mov_b32 v36, v16 :: v_dual_mov_b32 v37, v17 +; GCN-SDAG-NEXT: v_dual_mov_b32 v2, v16 :: v_dual_mov_b32 v3, v17 ; GCN-SDAG-NEXT: s_wait_xcnt 0x0 ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[12:13], v[12:13], v[12:13] ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[10:11], v[10:11], v[10:11] @@ -286,8 +288,8 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[8:9], v[8:9], v[8:9] ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[6:7], v[6:7], v[6:7] ; GCN-SDAG-NEXT: s_wait_loadcnt 0x0 -; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[50:51], v[2:3], v[2:3] -; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[48:49], v[0:1], v[0:1] +; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[50:51], v[36:37], v[36:37] +; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[48:49], v[34:35], v[34:35] ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[16:17], v[16:17], v[16:17] ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[14:15], 0xc8, v[14:15] ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[24:25], 0x64, v[24:25] @@ -298,8 +300,8 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[20:21], v[20:21], v[20:21] ; GCN-SDAG-NEXT: v_add_nc_u64_e32 v[18:19], v[18:19], v[18:19] ; GCN-SDAG-NEXT: s_clause 0x1 -; GCN-SDAG-NEXT: global_store_b128 v[52:53], v[34:37], off -; GCN-SDAG-NEXT: global_store_b128 v[54:55], v[0:3], off +; GCN-SDAG-NEXT: global_store_b128 v[52:53], v[0:3], off +; GCN-SDAG-NEXT: global_store_b128 v[54:55], v[34:37], off ; GCN-SDAG-NEXT: s_clause 0x7 ; GCN-SDAG-NEXT: global_store_b128 v[4:5], v[10:13], off offset:96 ; GCN-SDAG-NEXT: global_store_b128 v[4:5], v[6:9], off offset:112 @@ -309,7 +311,7 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-SDAG-NEXT: global_store_b128 v[4:5], v[18:21], off offset:48 ; GCN-SDAG-NEXT: global_store_b128 v[4:5], v[30:33], off ; GCN-SDAG-NEXT: global_store_b128 v[4:5], v[26:29], off offset:16 -; GCN-SDAG-NEXT: s_wait_xcnt 0x8 +; GCN-SDAG-NEXT: s_wait_xcnt 0x9 ; GCN-SDAG-NEXT: v_dual_mov_b32 v0, v32 :: v_dual_mov_b32 v1, v33 ; GCN-SDAG-NEXT: s_set_pc_i64 s[30:31] ; @@ -325,7 +327,7 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-GISEL-NEXT: global_load_b128 v[22:25], v[0:1], off offset:48 ; GCN-GISEL-NEXT: global_load_b128 v[26:29], v[0:1], off offset:96 ; GCN-GISEL-NEXT: global_load_b128 v[30:33], v[0:1], off offset:112 -; GCN-GISEL-NEXT: global_load_b128 v[0:3], v[0:1], off offset:64 +; GCN-GISEL-NEXT: global_load_b128 v[34:37], v[0:1], off offset:64 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[38:39], 0 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[48:49], 16 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[50:51], 32 @@ -333,7 +335,8 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-GISEL-NEXT: v_mov_b64_e32 v[66:67], 0x60 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[68:69], 0x70 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[54:55], 64 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[34:35], 0xc8 +; GCN-GISEL-NEXT: s_wait_xcnt 0x0 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[0:1], 0xc8 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[64:65], 0x50 ; GCN-GISEL-NEXT: s_wait_loadcnt 0x6 ; GCN-GISEL-NEXT: global_store_b128 v[38:39], v[10:13], off @@ -349,7 +352,7 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-GISEL-NEXT: global_store_b128 v[68:69], v[30:33], off ; GCN-GISEL-NEXT: s_wait_xcnt 0x5 ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[12:13], v[12:13], v[12:13] -; GCN-GISEL-NEXT: v_mov_b64_e32 v[36:37], v[8:9] +; GCN-GISEL-NEXT: v_mov_b64_e32 v[2:3], v[8:9] ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[10:11], v[10:11], v[10:11] ; GCN-GISEL-NEXT: s_wait_xcnt 0x4 ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[14:15], v[14:15], v[14:15] @@ -361,8 +364,8 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[22:23], v[22:23], v[22:23] ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[24:25], v[24:25], v[24:25] ; GCN-GISEL-NEXT: s_wait_loadcnt 0x0 -; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[48:49], v[0:1], v[0:1] -; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[50:51], v[2:3], v[2:3] +; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[48:49], v[34:35], v[34:35] +; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[50:51], v[36:37], v[36:37] ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[6:7], 0xc8, v[6:7] ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[8:9], v[8:9], v[8:9] ; GCN-GISEL-NEXT: s_wait_xcnt 0x1 @@ -372,8 +375,8 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[30:31], v[30:31], v[30:31] ; GCN-GISEL-NEXT: v_add_nc_u64_e32 v[32:33], v[32:33], v[32:33] ; GCN-GISEL-NEXT: s_clause 0x1 -; GCN-GISEL-NEXT: global_store_b128 v[54:55], v[0:3], off -; GCN-GISEL-NEXT: global_store_b128 v[64:65], v[34:37], off +; GCN-GISEL-NEXT: global_store_b128 v[54:55], v[34:37], off +; GCN-GISEL-NEXT: global_store_b128 v[64:65], v[0:3], off ; GCN-GISEL-NEXT: s_clause 0x7 ; GCN-GISEL-NEXT: global_store_b128 v[4:5], v[10:13], off ; GCN-GISEL-NEXT: global_store_b128 v[4:5], v[14:17], off offset:16 @@ -383,7 +386,7 @@ define i64 @test_v16i64_load_store(ptr addrspace(1) %ptr_a, ptr addrspace(1) %pt ; GCN-GISEL-NEXT: global_store_b128 v[4:5], v[6:9], off offset:80 ; GCN-GISEL-NEXT: global_store_b128 v[4:5], v[26:29], off offset:96 ; GCN-GISEL-NEXT: global_store_b128 v[4:5], v[30:33], off offset:112 -; GCN-GISEL-NEXT: s_wait_xcnt 0x9 +; GCN-GISEL-NEXT: s_wait_xcnt 0x8 ; GCN-GISEL-NEXT: v_dual_mov_b32 v0, v12 :: v_dual_mov_b32 v1, v13 ; GCN-GISEL-NEXT: s_set_pc_i64 s[30:31] %a = load <16 x i64>, ptr addrspace(1) %ptr_a, align 4 @@ -402,16 +405,17 @@ define amdgpu_kernel void @test_v7i16_load_store_kernel(ptr addrspace(1) %ptr1, ; GCN-SDAG-LABEL: test_v7i16_load_store_kernel: ; GCN-SDAG: ; %bb.0: ; GCN-SDAG-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GCN-SDAG-NEXT: v_and_b32_e32 v4, 0x3ff, v0 +; GCN-SDAG-NEXT: v_and_b32_e32 v8, 0x3ff, v0 ; GCN-SDAG-NEXT: s_wait_xcnt 0x0 ; GCN-SDAG-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 -; GCN-SDAG-NEXT: v_mov_b64_e32 v[8:9], 12 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[10:11], 8 ; GCN-SDAG-NEXT: v_mov_b64_e32 v[12:13], 0 ; GCN-SDAG-NEXT: s_wait_kmcnt 0x0 ; GCN-SDAG-NEXT: s_clause 0x1 -; GCN-SDAG-NEXT: global_load_b128 v[0:3], v4, s[0:1] scale_offset -; GCN-SDAG-NEXT: global_load_b128 v[4:7], v4, s[2:3] scale_offset +; GCN-SDAG-NEXT: global_load_b128 v[0:3], v8, s[0:1] scale_offset +; GCN-SDAG-NEXT: global_load_b128 v[4:7], v8, s[2:3] scale_offset +; GCN-SDAG-NEXT: s_wait_xcnt 0x0 +; GCN-SDAG-NEXT: v_mov_b64_e32 v[8:9], 12 ; GCN-SDAG-NEXT: s_wait_loadcnt 0x0 ; GCN-SDAG-NEXT: v_pk_add_u16 v3, v3, v7 ; GCN-SDAG-NEXT: v_pk_add_u16 v2, v2, v6 @@ -428,10 +432,9 @@ define amdgpu_kernel void @test_v7i16_load_store_kernel(ptr addrspace(1) %ptr1, ; GCN-GISEL-LABEL: test_v7i16_load_store_kernel: ; GCN-GISEL: ; %bb.0: ; GCN-GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GCN-GISEL-NEXT: v_and_b32_e32 v4, 0x3ff, v0 +; GCN-GISEL-NEXT: v_and_b32_e32 v8, 0x3ff, v0 ; GCN-GISEL-NEXT: s_wait_xcnt 0x0 ; GCN-GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 -; GCN-GISEL-NEXT: v_mov_b64_e32 v[8:9], 0 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[10:11], 2 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[12:13], 4 ; GCN-GISEL-NEXT: v_mov_b64_e32 v[14:15], 6 @@ -440,8 +443,10 @@ define amdgpu_kernel void @test_v7i16_load_store_kernel(ptr addrspace(1) %ptr1, ; GCN-GISEL-NEXT: v_mov_b64_e32 v[20:21], 12 ; GCN-GISEL-NEXT: s_wait_kmcnt 0x0 ; GCN-GISEL-NEXT: s_clause 0x1 -; GCN-GISEL-NEXT: global_load_b128 v[0:3], v4, s[0:1] scale_offset -; GCN-GISEL-NEXT: global_load_b128 v[4:7], v4, s[2:3] scale_offset +; GCN-GISEL-NEXT: global_load_b128 v[0:3], v8, s[0:1] scale_offset +; GCN-GISEL-NEXT: global_load_b128 v[4:7], v8, s[2:3] scale_offset +; GCN-GISEL-NEXT: s_wait_xcnt 0x0 +; GCN-GISEL-NEXT: v_mov_b64_e32 v[8:9], 0 ; GCN-GISEL-NEXT: s_wait_loadcnt 0x0 ; GCN-GISEL-NEXT: v_pk_add_u16 v0, v0, v4 ; GCN-GISEL-NEXT: v_pk_add_u16 v1, v1, v5 diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.id.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.id.ll index 90fcb51..fa97380 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.id.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.id.ll @@ -11,14 +11,11 @@ declare i32 @llvm.amdgcn.cluster.id.z() #0 define amdgpu_kernel void @test_cluster_id_x(ptr addrspace(1) %out) { ; CHECK-UNKNOWN-LABEL: test_cluster_id_x: ; CHECK-UNKNOWN: ; %bb.0: -; CHECK-UNKNOWN-NEXT: s_load_b64 s[0:1], s[0:1], 0x24 +; CHECK-UNKNOWN-NEXT: s_load_b64 s[2:3], s[0:1], 0x24 ; CHECK-UNKNOWN-NEXT: v_dual_mov_b32 v0, ttmp9 :: v_dual_mov_b32 v1, 0 ; CHECK-UNKNOWN-NEXT: s_wait_kmcnt 0x0 -; CHECK-UNKNOWN-NEXT: global_store_b32 v1, v0, s[0:1] +; CHECK-UNKNOWN-NEXT: global_store_b32 v1, v0, s[2:3] ; CHECK-UNKNOWN-NEXT: s_endpgm -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_X_EN: 1 -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Y_EN: 0 -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Z_EN: 0 ; ; CHECK-MESA3D-LABEL: test_cluster_id_x: ; CHECK-MESA3D: .amd_kernel_code_t @@ -68,7 +65,7 @@ define amdgpu_kernel void @test_cluster_id_x(ptr addrspace(1) %out) { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -98,14 +95,11 @@ define amdgpu_kernel void @test_cluster_id_x(ptr addrspace(1) %out) { ; ; CHECK-G-UNKNOWN-LABEL: test_cluster_id_x: ; CHECK-G-UNKNOWN: ; %bb.0: -; CHECK-G-UNKNOWN-NEXT: s_load_b64 s[0:1], s[0:1], 0x24 +; CHECK-G-UNKNOWN-NEXT: s_load_b64 s[2:3], s[0:1], 0x24 ; CHECK-G-UNKNOWN-NEXT: v_dual_mov_b32 v0, ttmp9 :: v_dual_mov_b32 v1, 0 ; CHECK-G-UNKNOWN-NEXT: s_wait_kmcnt 0x0 -; CHECK-G-UNKNOWN-NEXT: global_store_b32 v1, v0, s[0:1] +; CHECK-G-UNKNOWN-NEXT: global_store_b32 v1, v0, s[2:3] ; CHECK-G-UNKNOWN-NEXT: s_endpgm -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_X_EN: 1 -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Y_EN: 0 -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Z_EN: 0 ; ; CHECK-G-MESA3D-LABEL: test_cluster_id_x: ; CHECK-G-MESA3D: .amd_kernel_code_t @@ -155,7 +149,7 @@ define amdgpu_kernel void @test_cluster_id_x(ptr addrspace(1) %out) { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -190,14 +184,11 @@ define amdgpu_kernel void @test_cluster_id_x(ptr addrspace(1) %out) { define amdgpu_kernel void @test_cluster_id_y(ptr addrspace(1) %out) #1 { ; CHECK-UNKNOWN-LABEL: test_cluster_id_y: ; CHECK-UNKNOWN: ; %bb.0: -; CHECK-UNKNOWN-NEXT: s_load_b64 s[0:1], s[0:1], 0x24 +; CHECK-UNKNOWN-NEXT: s_load_b64 s[2:3], s[0:1], 0x24 ; CHECK-UNKNOWN-NEXT: v_dual_mov_b32 v0, ttmp7 :: v_dual_mov_b32 v1, 0 ; CHECK-UNKNOWN-NEXT: s_wait_kmcnt 0x0 -; CHECK-UNKNOWN-NEXT: global_store_b32 v1, v0, s[0:1] +; CHECK-UNKNOWN-NEXT: global_store_b32 v1, v0, s[2:3] ; CHECK-UNKNOWN-NEXT: s_endpgm -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_X_EN: 1 -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Y_EN: 1 -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Z_EN: 0 ; ; CHECK-MESA3D-LABEL: test_cluster_id_y: ; CHECK-MESA3D: .amd_kernel_code_t @@ -247,7 +238,7 @@ define amdgpu_kernel void @test_cluster_id_y(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -277,14 +268,11 @@ define amdgpu_kernel void @test_cluster_id_y(ptr addrspace(1) %out) #1 { ; ; CHECK-G-UNKNOWN-LABEL: test_cluster_id_y: ; CHECK-G-UNKNOWN: ; %bb.0: -; CHECK-G-UNKNOWN-NEXT: s_load_b64 s[0:1], s[0:1], 0x24 +; CHECK-G-UNKNOWN-NEXT: s_load_b64 s[2:3], s[0:1], 0x24 ; CHECK-G-UNKNOWN-NEXT: v_dual_mov_b32 v0, ttmp7 :: v_dual_mov_b32 v1, 0 ; CHECK-G-UNKNOWN-NEXT: s_wait_kmcnt 0x0 -; CHECK-G-UNKNOWN-NEXT: global_store_b32 v1, v0, s[0:1] +; CHECK-G-UNKNOWN-NEXT: global_store_b32 v1, v0, s[2:3] ; CHECK-G-UNKNOWN-NEXT: s_endpgm -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_X_EN: 1 -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Y_EN: 1 -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Z_EN: 0 ; ; CHECK-G-MESA3D-LABEL: test_cluster_id_y: ; CHECK-G-MESA3D: .amd_kernel_code_t @@ -334,7 +322,7 @@ define amdgpu_kernel void @test_cluster_id_y(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -369,16 +357,14 @@ define amdgpu_kernel void @test_cluster_id_y(ptr addrspace(1) %out) #1 { define amdgpu_kernel void @test_cluster_id_z(ptr addrspace(1) %out) #1 { ; CHECK-UNKNOWN-LABEL: test_cluster_id_z: ; CHECK-UNKNOWN: ; %bb.0: -; CHECK-UNKNOWN-NEXT: s_load_b64 s[0:1], s[0:1], 0x24 -; CHECK-UNKNOWN-NEXT: s_lshr_b32 s2, ttmp7, 16 +; CHECK-UNKNOWN-NEXT: s_load_b64 s[2:3], s[0:1], 0x24 +; CHECK-UNKNOWN-NEXT: s_wait_xcnt 0x0 +; CHECK-UNKNOWN-NEXT: s_lshr_b32 s0, ttmp7, 16 ; CHECK-UNKNOWN-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; CHECK-UNKNOWN-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s2 +; CHECK-UNKNOWN-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s0 ; CHECK-UNKNOWN-NEXT: s_wait_kmcnt 0x0 -; CHECK-UNKNOWN-NEXT: global_store_b32 v0, v1, s[0:1] +; CHECK-UNKNOWN-NEXT: global_store_b32 v0, v1, s[2:3] ; CHECK-UNKNOWN-NEXT: s_endpgm -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_X_EN: 1 -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Y_EN: 0 -; CHECK-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Z_EN: 1 ; ; CHECK-MESA3D-LABEL: test_cluster_id_z: ; CHECK-MESA3D: .amd_kernel_code_t @@ -428,7 +414,7 @@ define amdgpu_kernel void @test_cluster_id_z(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -460,16 +446,14 @@ define amdgpu_kernel void @test_cluster_id_z(ptr addrspace(1) %out) #1 { ; ; CHECK-G-UNKNOWN-LABEL: test_cluster_id_z: ; CHECK-G-UNKNOWN: ; %bb.0: -; CHECK-G-UNKNOWN-NEXT: s_load_b64 s[0:1], s[0:1], 0x24 -; CHECK-G-UNKNOWN-NEXT: s_lshr_b32 s2, ttmp7, 16 +; CHECK-G-UNKNOWN-NEXT: s_load_b64 s[2:3], s[0:1], 0x24 +; CHECK-G-UNKNOWN-NEXT: s_wait_xcnt 0x0 +; CHECK-G-UNKNOWN-NEXT: s_lshr_b32 s0, ttmp7, 16 ; CHECK-G-UNKNOWN-NEXT: s_delay_alu instid0(SALU_CYCLE_1) -; CHECK-G-UNKNOWN-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, s2 +; CHECK-G-UNKNOWN-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, s0 ; CHECK-G-UNKNOWN-NEXT: s_wait_kmcnt 0x0 -; CHECK-G-UNKNOWN-NEXT: global_store_b32 v1, v0, s[0:1] +; CHECK-G-UNKNOWN-NEXT: global_store_b32 v1, v0, s[2:3] ; CHECK-G-UNKNOWN-NEXT: s_endpgm -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_X_EN: 1 -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Y_EN: 0 -; CHECK-G-UNKNOWN: COMPUTE_PGM_RSRC2:TGID_Z_EN: 1 ; ; CHECK-G-MESA3D-LABEL: test_cluster_id_z: ; CHECK-G-MESA3D: .amd_kernel_code_t @@ -519,7 +503,7 @@ define amdgpu_kernel void @test_cluster_id_z(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.id.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.id.ll index aa3b7b3..3ef84a3 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.id.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.id.ll @@ -67,7 +67,7 @@ define amdgpu_kernel void @test_workgroup_id_x(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -155,7 +155,7 @@ define amdgpu_kernel void @test_workgroup_id_x(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -246,7 +246,7 @@ define amdgpu_kernel void @test_workgroup_id_x_optimized(ptr addrspace(1) %out) ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -330,7 +330,7 @@ define amdgpu_kernel void @test_workgroup_id_x_optimized(ptr addrspace(1) %out) ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -421,7 +421,7 @@ define amdgpu_kernel void @test_workgroup_id_y(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -509,7 +509,7 @@ define amdgpu_kernel void @test_workgroup_id_y(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -600,7 +600,7 @@ define amdgpu_kernel void @test_workgroup_id_y_optimized(ptr addrspace(1) %out) ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -684,7 +684,7 @@ define amdgpu_kernel void @test_workgroup_id_y_optimized(ptr addrspace(1) %out) ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -775,7 +775,7 @@ define amdgpu_kernel void @test_workgroup_id_z(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -863,7 +863,7 @@ define amdgpu_kernel void @test_workgroup_id_z(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -956,7 +956,7 @@ define amdgpu_kernel void @test_workgroup_flat_id(ptr addrspace(1) %out) { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -1044,7 +1044,7 @@ define amdgpu_kernel void @test_workgroup_flat_id(ptr addrspace(1) %out) { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -1135,7 +1135,7 @@ define amdgpu_kernel void @test_workgroup_id_z_optimized(ptr addrspace(1) %out) ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -1219,7 +1219,7 @@ define amdgpu_kernel void @test_workgroup_id_z_optimized(ptr addrspace(1) %out) ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.flat.id.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.flat.id.ll index afe37e3..b8ff9e5 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.flat.id.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.flat.id.ll @@ -65,7 +65,7 @@ define amdgpu_kernel void @test_workgroup_max_flat_id(ptr addrspace(1) %out) #1 ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -153,7 +153,7 @@ define amdgpu_kernel void @test_workgroup_max_flat_id(ptr addrspace(1) %out) #1 ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.id.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.id.ll index 7ea4fa5..9bca696 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.id.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cluster.workgroup.max.id.ll @@ -67,7 +67,7 @@ define amdgpu_kernel void @test_workgroup_max_id_x(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -155,7 +155,7 @@ define amdgpu_kernel void @test_workgroup_max_id_x(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -246,7 +246,7 @@ define amdgpu_kernel void @test_workgroup_max_id_x_optimized(ptr addrspace(1) %o ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -330,7 +330,7 @@ define amdgpu_kernel void @test_workgroup_max_id_x_optimized(ptr addrspace(1) %o ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -421,7 +421,7 @@ define amdgpu_kernel void @test_workgroup_max_id_y(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -509,7 +509,7 @@ define amdgpu_kernel void @test_workgroup_max_id_y(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -600,7 +600,7 @@ define amdgpu_kernel void @test_workgroup_max_id_y_optimized(ptr addrspace(1) %o ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -684,7 +684,7 @@ define amdgpu_kernel void @test_workgroup_max_id_y_optimized(ptr addrspace(1) %o ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -775,7 +775,7 @@ define amdgpu_kernel void @test_workgroup_max_id_z(ptr addrspace(1) %out) #1 { ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -863,7 +863,7 @@ define amdgpu_kernel void @test_workgroup_max_id_z(ptr addrspace(1) %out) #1 { ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -954,7 +954,7 @@ define amdgpu_kernel void @test_workgroup_max_id_z_optimized(ptr addrspace(1) %o ; CHECK-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-MESA3D-NEXT: gds_segment_byte_size = 0 @@ -1038,7 +1038,7 @@ define amdgpu_kernel void @test_workgroup_max_id_z_optimized(ptr addrspace(1) %o ; CHECK-G-MESA3D-NEXT: is_ptr64 = 1 ; CHECK-G-MESA3D-NEXT: is_dynamic_callstack = 0 ; CHECK-G-MESA3D-NEXT: is_debug_enabled = 0 -; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 0 +; CHECK-G-MESA3D-NEXT: is_xnack_enabled = 1 ; CHECK-G-MESA3D-NEXT: workitem_private_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: workgroup_group_segment_byte_size = 0 ; CHECK-G-MESA3D-NEXT: gds_segment_byte_size = 0 diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll index 56215ca..67d0410 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll @@ -59,21 +59,20 @@ define amdgpu_kernel void @is_private_vgpr(ptr addrspace(1) %ptr.ptr) { ; GFX9-NEXT: global_store_dword v[0:1], v0, off ; GFX9-NEXT: s_endpgm ; -; GFX1250-SDAG-LABEL: is_private_vgpr: -; GFX1250-SDAG: ; %bb.0: -; GFX1250-SDAG-NEXT: s_load_b64 s[0:1], s[4:5], 0x0 -; GFX1250-SDAG-NEXT: v_and_b32_e32 v0, 0x3ff, v0 -; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 -; GFX1250-SDAG-NEXT: global_load_b64 v[0:1], v0, s[0:1] scale_offset scope:SCOPE_SYS -; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x0 -; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 -; GFX1250-SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX1250-SDAG-NEXT: v_xor_b32_e32 v0, s0, v1 -; GFX1250-SDAG-NEXT: v_cmp_gt_u32_e32 vcc_lo, 0x4000000, v0 -; GFX1250-SDAG-NEXT: v_cndmask_b32_e64 v0, 0, 1, vcc_lo -; GFX1250-SDAG-NEXT: global_store_b32 v[0:1], v0, off -; GFX1250-SDAG-NEXT: s_endpgm +; GFX1250-LABEL: is_private_vgpr: +; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_load_b64 s[0:1], s[4:5], 0x0 +; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 +; GFX1250-NEXT: global_load_b64 v[0:1], v0, s[0:1] scale_offset scope:SCOPE_SYS +; GFX1250-NEXT: s_wait_loadcnt 0x0 +; GFX1250-NEXT: s_wait_xcnt 0x0 +; GFX1250-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v1 +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1250-NEXT: v_cmp_gt_u32_e32 vcc_lo, 0x4000000, v0 +; GFX1250-NEXT: v_cndmask_b32_e64 v0, 0, 1, vcc_lo +; GFX1250-NEXT: global_store_b32 v[0:1], v0, off +; GFX1250-NEXT: s_endpgm ; ; CI-GISEL-LABEL: is_private_vgpr: ; CI-GISEL: ; %bb.0: @@ -122,22 +121,6 @@ define amdgpu_kernel void @is_private_vgpr(ptr addrspace(1) %ptr.ptr) { ; GFX11-NEXT: v_cndmask_b32_e64 v0, 0, 1, vcc_lo ; GFX11-NEXT: global_store_b32 v[0:1], v0, off ; GFX11-NEXT: s_endpgm -; -; GFX1250-GISEL-LABEL: is_private_vgpr: -; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x0 -; GFX1250-GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v2, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 -; GFX1250-GISEL-NEXT: global_load_b64 v[0:1], v0, s[0:1] scale_offset scope:SCOPE_SYS -; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x0 -; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 -; GFX1250-GISEL-NEXT: v_xor_b32_e32 v0, v1, v2 -; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-GISEL-NEXT: v_cmp_gt_u32_e32 vcc_lo, 0x4000000, v0 -; GFX1250-GISEL-NEXT: v_cndmask_b32_e64 v0, 0, 1, vcc_lo -; GFX1250-GISEL-NEXT: global_store_b32 v[0:1], v0, off -; GFX1250-GISEL-NEXT: s_endpgm %id = call i32 @llvm.amdgcn.workitem.id.x() %gep = getelementptr inbounds ptr, ptr addrspace(1) %ptr.ptr, i32 %id %ptr = load volatile ptr, ptr addrspace(1) %gep @@ -206,9 +189,8 @@ define amdgpu_kernel void @is_private_sgpr(ptr %ptr) { ; GFX1250-SDAG-LABEL: is_private_sgpr: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: s_load_b32 s0, s[4:5], 0x4 -; GFX1250-SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 -; GFX1250-SDAG-NEXT: s_xor_b32 s0, s0, s1 +; GFX1250-SDAG-NEXT: s_xor_b32 s0, s0, src_flat_scratch_base_hi ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; GFX1250-SDAG-NEXT: s_cmp_lt_u32 s0, 0x4000000 ; GFX1250-SDAG-NEXT: s_cselect_b32 s0, -1, 0 @@ -285,9 +267,8 @@ define amdgpu_kernel void @is_private_sgpr(ptr %ptr) { ; GFX1250-GISEL: ; %bb.0: ; GFX1250-GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x0 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 -; GFX1250-GISEL-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_xor_b32 s0, s1, s0 +; GFX1250-GISEL-NEXT: s_xor_b32 s0, s1, src_flat_scratch_base_hi +; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-GISEL-NEXT: s_cmp_ge_u32 s0, 0x4000000 ; GFX1250-GISEL-NEXT: s_cbranch_scc1 .LBB1_2 ; GFX1250-GISEL-NEXT: ; %bb.1: ; %bb0 @@ -311,5 +292,4 @@ bb1: ; CI: {{.*}} ; GFX10-GISEL: {{.*}} ; GFX11-GISEL: {{.*}} -; GFX1250: {{.*}} ; SI-SDAG: {{.*}} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.permlane.gfx1250.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.permlane.gfx1250.ll index 4f7bbf8..42a50bb 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.permlane.gfx1250.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.permlane.gfx1250.ll @@ -5,13 +5,13 @@ define amdgpu_kernel void @v_permlane_bcast_b32_vss(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) { ; GFX1250-LABEL: v_permlane_bcast_b32_vss: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b32 s4, s[4:5], 0x34 +; GFX1250-NEXT: s_load_b32 s6, s[4:5], 0x34 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, s2 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-NEXT: v_permlane_bcast_b32 v0, v0, s3, s4 +; GFX1250-NEXT: v_permlane_bcast_b32 v0, v0, s3, s6 ; GFX1250-NEXT: global_store_b32 v1, v0, s[0:1] ; GFX1250-NEXT: s_endpgm %v = call i32 @llvm.amdgcn.permlane.bcast(i32 %src0, i32 %src1, i32 %src2) @@ -92,13 +92,13 @@ define amdgpu_kernel void @v_permlane_bcast_b32_vvv(ptr addrspace(1) %out, i32 % define amdgpu_kernel void @v_permlane_down_b32_vss(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) { ; GFX1250-LABEL: v_permlane_down_b32_vss: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b32 s4, s[4:5], 0x34 +; GFX1250-NEXT: s_load_b32 s6, s[4:5], 0x34 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, s2 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-NEXT: v_permlane_down_b32 v0, v0, s3, s4 +; GFX1250-NEXT: v_permlane_down_b32 v0, v0, s3, s6 ; GFX1250-NEXT: global_store_b32 v1, v0, s[0:1] ; GFX1250-NEXT: s_endpgm %v = call i32 @llvm.amdgcn.permlane.down(i32 %src0, i32 %src1, i32 %src2) @@ -179,13 +179,13 @@ define amdgpu_kernel void @v_permlane_down_b32_vvv(ptr addrspace(1) %out, i32 %s define amdgpu_kernel void @v_permlane_up_b32_vss(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) { ; GFX1250-LABEL: v_permlane_up_b32_vss: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b32 s4, s[4:5], 0x34 +; GFX1250-NEXT: s_load_b32 s6, s[4:5], 0x34 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, s2 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-NEXT: v_permlane_up_b32 v0, v0, s3, s4 +; GFX1250-NEXT: v_permlane_up_b32 v0, v0, s3, s6 ; GFX1250-NEXT: global_store_b32 v1, v0, s[0:1] ; GFX1250-NEXT: s_endpgm %v = call i32 @llvm.amdgcn.permlane.up(i32 %src0, i32 %src1, i32 %src2) @@ -266,13 +266,13 @@ define amdgpu_kernel void @v_permlane_up_b32_vvv(ptr addrspace(1) %out, i32 %src define amdgpu_kernel void @v_permlane_xor_b32_vss(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) { ; GFX1250-LABEL: v_permlane_xor_b32_vss: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b32 s4, s[4:5], 0x34 +; GFX1250-NEXT: s_load_b32 s6, s[4:5], 0x34 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, s2 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) -; GFX1250-NEXT: v_permlane_xor_b32 v0, v0, s3, s4 +; GFX1250-NEXT: v_permlane_xor_b32 v0, v0, s3, s6 ; GFX1250-NEXT: global_store_b32 v1, v0, s[0:1] ; GFX1250-NEXT: s_endpgm %v = call i32 @llvm.amdgcn.permlane.xor(i32 %src0, i32 %src1, i32 %src2) diff --git a/llvm/test/CodeGen/AMDGPU/load-store-opt-scale-offset.mir b/llvm/test/CodeGen/AMDGPU/load-store-opt-scale-offset.mir index 76e2092..abcae69 100644 --- a/llvm/test/CodeGen/AMDGPU/load-store-opt-scale-offset.mir +++ b/llvm/test/CodeGen/AMDGPU/load-store-opt-scale-offset.mir @@ -69,9 +69,9 @@ body: | bb.0: ; GCN-LABEL: name: merge_s_load_x1_x1_imm_no_scale_offset ; GCN: [[DEF:%[0-9]+]]:sgpr_64 = IMPLICIT_DEF - ; GCN-NEXT: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[DEF]], 0, 0 :: (dereferenceable invariant load (s64), align 4) - ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_32_xm0_xexec = COPY [[S_LOAD_DWORDX2_IMM]].sub0 - ; GCN-NEXT: [[COPY1:%[0-9]+]]:sreg_32_xm0_xexec = COPY killed [[S_LOAD_DWORDX2_IMM]].sub1 + ; GCN-NEXT: early-clobber %3:sreg_64_xexec = S_LOAD_DWORDX2_IMM_ec [[DEF]], 0, 0 :: (dereferenceable invariant load (s64), align 4) + ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_32_xm0_xexec = COPY %3.sub0 + ; GCN-NEXT: [[COPY1:%[0-9]+]]:sreg_32_xm0_xexec = COPY killed %3.sub1 %0:sgpr_64 = IMPLICIT_DEF %1:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %0:sgpr_64, 0, 0 :: (dereferenceable invariant load (s32)) %2:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %0:sgpr_64, 4, 0 :: (dereferenceable invariant load (s32)) diff --git a/llvm/test/CodeGen/AMDGPU/loop-prefetch-data.ll b/llvm/test/CodeGen/AMDGPU/loop-prefetch-data.ll index 1e6b77e..4ad161c 100644 --- a/llvm/test/CodeGen/AMDGPU/loop-prefetch-data.ll +++ b/llvm/test/CodeGen/AMDGPU/loop-prefetch-data.ll @@ -471,13 +471,13 @@ define amdgpu_kernel void @copy_flat_divergent(ptr nocapture %d, ptr nocapture r ; GFX1250-NEXT: s_cmp_eq_u32 s0, 0 ; GFX1250-NEXT: s_cbranch_scc1 .LBB4_3 ; GFX1250-NEXT: ; %bb.1: ; %for.body.preheader -; GFX1250-NEXT: s_load_b128 s[4:7], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b128 s[8:11], s[4:5], 0x24 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_lshlrev_b32 v0, 4, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_add_nc_u64_e32 v[2:3], s[6:7], v[0:1] -; GFX1250-NEXT: v_add_nc_u64_e32 v[0:1], s[4:5], v[0:1] +; GFX1250-NEXT: v_add_nc_u64_e32 v[2:3], s[10:11], v[0:1] +; GFX1250-NEXT: v_add_nc_u64_e32 v[0:1], s[8:9], v[0:1] ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) ; GFX1250-NEXT: v_add_nc_u64_e32 v[2:3], 0xb0, v[2:3] ; GFX1250-NEXT: .LBB4_2: ; %for.body @@ -602,13 +602,13 @@ define amdgpu_kernel void @copy_global_divergent(ptr addrspace(1) nocapture %d, ; GFX1250-NEXT: s_cmp_eq_u32 s0, 0 ; GFX1250-NEXT: s_cbranch_scc1 .LBB5_3 ; GFX1250-NEXT: ; %bb.1: ; %for.body.preheader -; GFX1250-NEXT: s_load_b128 s[4:7], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b128 s[8:11], s[4:5], 0x24 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_lshlrev_b32 v0, 4, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_add_nc_u64_e32 v[2:3], s[6:7], v[0:1] -; GFX1250-NEXT: v_add_nc_u64_e32 v[0:1], s[4:5], v[0:1] +; GFX1250-NEXT: v_add_nc_u64_e32 v[2:3], s[10:11], v[0:1] +; GFX1250-NEXT: v_add_nc_u64_e32 v[0:1], s[8:9], v[0:1] ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) ; GFX1250-NEXT: v_add_nc_u64_e32 v[2:3], 0xb0, v[2:3] ; GFX1250-NEXT: .LBB5_2: ; %for.body diff --git a/llvm/test/CodeGen/AMDGPU/mad_64_32.ll b/llvm/test/CodeGen/AMDGPU/mad_64_32.ll index dbcd370..08ec0c8 100644 --- a/llvm/test/CodeGen/AMDGPU/mad_64_32.ll +++ b/llvm/test/CodeGen/AMDGPU/mad_64_32.ll @@ -1117,18 +1117,19 @@ define amdgpu_kernel void @mad_i64_i32_uniform(ptr addrspace(1) %out, i32 %arg0, ; ; GFX1250-LABEL: mad_i64_i32_uniform: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 -; GFX1250-NEXT: s_mov_b32 s7, 0 +; GFX1250-NEXT: s_mov_b32 s5, 0 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mov_b32 s6, s2 +; GFX1250-NEXT: s_mov_b32 s4, s2 ; GFX1250-NEXT: s_mov_b32 s2, s3 -; GFX1250-NEXT: s_mov_b32 s3, s7 +; GFX1250-NEXT: s_mov_b32 s3, s5 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: s_mul_u64 s[2:3], s[6:7], s[2:3] -; GFX1250-NEXT: s_add_nc_u64 s[2:3], s[2:3], s[4:5] +; GFX1250-NEXT: s_mul_u64 s[2:3], s[4:5], s[2:3] +; GFX1250-NEXT: s_add_nc_u64 s[2:3], s[2:3], s[6:7] ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] diff --git a/llvm/test/CodeGen/AMDGPU/mai-hazards.mir b/llvm/test/CodeGen/AMDGPU/mai-hazards.mir index 61f2629..c19d5a6 100644 --- a/llvm/test/CodeGen/AMDGPU/mai-hazards.mir +++ b/llvm/test/CodeGen/AMDGPU/mai-hazards.mir @@ -33,7 +33,7 @@ name: asm_write_vgpr_accvgpr_write_read body: | bb.0: - INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 2228234 /* regdef:VGPR_32 */, def $vgpr0 + INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 2031626 /* regdef:VGPR_32 */, def $vgpr0 $agpr0 = V_ACCVGPR_WRITE_B32_e64 killed $vgpr0, implicit $exec ... @@ -47,7 +47,7 @@ name: asm_write_vgpr_accvgpr_write_read_partialnop body: | bb.0: - INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 2228234 /* regdef:VGPR_32 */, def $vgpr0 + INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 2031626 /* regdef:VGPR_32 */, def $vgpr0 S_NOP 0 $agpr0 = V_ACCVGPR_WRITE_B32_e64 killed $vgpr0, implicit $exec ... @@ -60,7 +60,7 @@ name: asm_write_vgpr_accvgpr_write_read_otherreg body: | bb.0: liveins: $vgpr0 - INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 2228234 /* regdef:VGPR_32 */, def $vgpr1 + INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 2031626 /* regdef:VGPR_32 */, def $vgpr1 $agpr0 = V_ACCVGPR_WRITE_B32_e64 killed $vgpr0, implicit $exec ... diff --git a/llvm/test/CodeGen/AMDGPU/max.ll b/llvm/test/CodeGen/AMDGPU/max.ll index fef9a9a..ae08054 100644 --- a/llvm/test/CodeGen/AMDGPU/max.ll +++ b/llvm/test/CodeGen/AMDGPU/max.ll @@ -257,16 +257,15 @@ define amdgpu_kernel void @v_test_imax_sge_i8(ptr addrspace(1) %out, ptr addrspa ; ; GFX1250-LABEL: v_test_imax_sge_i8: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v0, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_load_i8 s2, s[2:3], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_i8 s3, s[4:5], 0x0 +; GFX1250-NEXT: s_load_i8 s4, s[2:3], 0x0 +; GFX1250-NEXT: s_load_i8 s5, s[6:7], 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_max_i32 s2, s2, s3 +; GFX1250-NEXT: s_max_i32 s2, s4, s5 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: v_mov_b32_e32 v1, s2 ; GFX1250-NEXT: global_store_b8 v0, v1, s[0:1] @@ -701,16 +700,15 @@ define amdgpu_kernel void @v_test_umax_uge_i8(ptr addrspace(1) %out, ptr addrspa ; ; GFX1250-LABEL: v_test_umax_uge_i8: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v0, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_load_u8 s2, s[2:3], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_u8 s3, s[4:5], 0x0 +; GFX1250-NEXT: s_load_u8 s4, s[2:3], 0x0 +; GFX1250-NEXT: s_load_u8 s5, s[6:7], 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_max_u32 s2, s2, s3 +; GFX1250-NEXT: s_max_u32 s2, s4, s5 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX1250-NEXT: v_mov_b32_e32 v1, s2 ; GFX1250-NEXT: global_store_b8 v0, v1, s[0:1] @@ -777,13 +775,12 @@ define amdgpu_kernel void @v_test_umax_ugt_i32(ptr addrspace(1) %out, ptr addrsp ; GFX1250-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: global_load_b32 v0, v0, s[0:1] scale_offset -; GFX1250-NEXT: s_load_b32 s2, s[0:1], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b32 s6, s[0:1], 0x0 +; GFX1250-NEXT: s_load_b64 s[2:3], s[4:5], 0x24 ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_max_u32_e32 v0, s2, v0 -; GFX1250-NEXT: global_store_b32 v1, v0, s[0:1] +; GFX1250-NEXT: v_max_u32_e32 v0, s6, v0 +; GFX1250-NEXT: global_store_b32 v1, v0, s[2:3] ; GFX1250-NEXT: s_endpgm ; ; EG-LABEL: v_test_umax_ugt_i32: @@ -1122,12 +1119,12 @@ define amdgpu_kernel void @test_umax_ugt_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_umax_ugt_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_max_u64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_max_u64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm ; @@ -1175,12 +1172,12 @@ define amdgpu_kernel void @test_umax_uge_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_umax_uge_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_max_u64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_max_u64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm ; @@ -1228,12 +1225,12 @@ define amdgpu_kernel void @test_imax_sgt_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_imax_sgt_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_max_i64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_max_i64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm ; @@ -1281,12 +1278,12 @@ define amdgpu_kernel void @test_imax_sge_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_imax_sge_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_max_i64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_max_i64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm ; diff --git a/llvm/test/CodeGen/AMDGPU/min.ll b/llvm/test/CodeGen/AMDGPU/min.ll index 311527d..6a3d31f 100644 --- a/llvm/test/CodeGen/AMDGPU/min.ll +++ b/llvm/test/CodeGen/AMDGPU/min.ll @@ -131,14 +131,14 @@ define amdgpu_kernel void @v_test_imin_sle_i32(ptr addrspace(1) %out, ptr addrsp ; ; GFX1250-LABEL: v_test_imin_sle_i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b32 v1, v0, s[2:3] scale_offset -; GFX1250-NEXT: global_load_b32 v2, v0, s[4:5] scale_offset +; GFX1250-NEXT: global_load_b32 v2, v0, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_min_i32_e32 v1, v1, v2 ; GFX1250-NEXT: global_store_b32 v0, v1, s[0:1] scale_offset @@ -1172,14 +1172,14 @@ define amdgpu_kernel void @s_test_imin_sle_v4i16(ptr addrspace(1) %out, <4 x i16 ; ; GFX1250-LABEL: s_test_imin_sle_v4i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x8 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x0 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x0 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_pk_min_i16 v1, s1, s3 ; GFX1250-NEXT: v_pk_min_i16 v0, s0, s2 -; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[4:5] +; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[6:7] ; GFX1250-NEXT: s_endpgm %cmp = icmp sle <4 x i16> %a, %b %val = select <4 x i1> %cmp, <4 x i16> %a, <4 x i16> %b @@ -1307,14 +1307,14 @@ define amdgpu_kernel void @v_test_imin_slt_i32(ptr addrspace(1) %out, ptr addrsp ; ; GFX1250-LABEL: v_test_imin_slt_i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b32 v1, v0, s[2:3] scale_offset -; GFX1250-NEXT: global_load_b32 v2, v0, s[4:5] scale_offset +; GFX1250-NEXT: global_load_b32 v2, v0, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_min_i32_e32 v1, v1, v2 ; GFX1250-NEXT: global_store_b32 v0, v1, s[0:1] scale_offset @@ -1484,14 +1484,14 @@ define amdgpu_kernel void @v_test_imin_slt_i16(ptr addrspace(1) %out, ptr addrsp ; ; GFX1250-LABEL: v_test_imin_slt_i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_u16 v1, v0, s[2:3] scale_offset -; GFX1250-NEXT: global_load_u16 v2, v0, s[4:5] scale_offset +; GFX1250-NEXT: global_load_u16 v2, v0, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_min_i16 v1, v1, v2 ; GFX1250-NEXT: global_store_b16 v0, v1, s[0:1] scale_offset @@ -1686,16 +1686,16 @@ define amdgpu_kernel void @s_test_imin_slt_v2i32(ptr addrspace(1) %out, <2 x i32 ; ; GFX1250-LABEL: s_test_imin_slt_v2i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x8 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x0 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x0 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_min_i32 s0, s0, s2 ; GFX1250-NEXT: s_min_i32 s1, s1, s3 ; GFX1250-NEXT: v_mov_b32_e32 v0, s0 ; GFX1250-NEXT: v_mov_b32_e32 v1, s1 -; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[4:5] +; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[6:7] ; GFX1250-NEXT: s_endpgm %cmp = icmp slt <2 x i32> %a, %b %val = select <2 x i1> %cmp, <2 x i32> %a, <2 x i32> %b @@ -2011,14 +2011,14 @@ define amdgpu_kernel void @v_test_umin_ule_i32(ptr addrspace(1) %out, ptr addrsp ; ; GFX1250-LABEL: v_test_umin_ule_i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b32 v1, v0, s[2:3] scale_offset -; GFX1250-NEXT: global_load_b32 v2, v0, s[4:5] scale_offset +; GFX1250-NEXT: global_load_b32 v2, v0, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_min_u32_e32 v1, v1, v2 ; GFX1250-NEXT: global_store_b32 v0, v1, s[0:1] scale_offset @@ -2171,16 +2171,16 @@ define amdgpu_kernel void @v_test_umin_ule_v3i32(ptr addrspace(1) %out, ptr addr ; ; GFX1250-LABEL: v_test_umin_ule_v3i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_lshlrev_b32_e32 v3, 4, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b96 v[0:2], v3, s[2:3] -; GFX1250-NEXT: global_load_b96 v[4:6], v3, s[4:5] +; GFX1250-NEXT: global_load_b96 v[4:6], v3, s[6:7] ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_min_u32_e32 v2, v2, v6 ; GFX1250-NEXT: v_min_u32_e32 v1, v1, v5 @@ -2374,14 +2374,14 @@ define amdgpu_kernel void @v_test_umin_ule_v3i16(ptr addrspace(1) %out, ptr addr ; ; GFX1250-LABEL: v_test_umin_ule_v3i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v4, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b64 v[0:1], v4, s[2:3] scale_offset -; GFX1250-NEXT: global_load_b64 v[2:3], v4, s[4:5] scale_offset +; GFX1250-NEXT: global_load_b64 v[2:3], v4, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_xcnt 0x0 ; GFX1250-NEXT: v_lshlrev_b32_e32 v4, 3, v4 ; GFX1250-NEXT: s_wait_loadcnt 0x0 @@ -2611,14 +2611,14 @@ define amdgpu_kernel void @v_test_umin_ult_i32(ptr addrspace(1) %out, ptr addrsp ; ; GFX1250-LABEL: v_test_umin_ult_i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b32 v1, v0, s[2:3] scale_offset -; GFX1250-NEXT: global_load_b32 v2, v0, s[4:5] scale_offset +; GFX1250-NEXT: global_load_b32 v2, v0, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_min_u32_e32 v1, v1, v2 ; GFX1250-NEXT: global_store_b32 v0, v1, s[0:1] scale_offset @@ -2771,14 +2771,14 @@ define amdgpu_kernel void @v_test_umin_ult_i8(ptr addrspace(1) %out, ptr addrspa ; ; GFX1250-LABEL: v_test_umin_ult_i8: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_u8 v1, v0, s[2:3] -; GFX1250-NEXT: global_load_u8 v2, v0, s[4:5] +; GFX1250-NEXT: global_load_u8 v2, v0, s[6:7] ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_min_u16 v1, v1, v2 ; GFX1250-NEXT: global_store_b8 v0, v1, s[0:1] @@ -3023,23 +3023,22 @@ define amdgpu_kernel void @v_test_umin_ult_i32_multi_use(ptr addrspace(1) %out0, ; ; GFX1250-LABEL: v_test_umin_ult_i32_multi_use: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x0 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x0 ; GFX1250-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_load_b32 s4, s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b32 s5, s[6:7], 0x0 +; GFX1250-NEXT: s_load_b32 s0, s[12:13], 0x0 +; GFX1250-NEXT: s_load_b32 s1, s[14:15], 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_cmp_lt_u32 s4, s5 -; GFX1250-NEXT: s_cselect_b32 s6, -1, 0 +; GFX1250-NEXT: s_cmp_lt_u32 s0, s1 +; GFX1250-NEXT: s_cselect_b32 s2, -1, 0 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) -; GFX1250-NEXT: v_cndmask_b32_e64 v0, 0, 1, s6 -; GFX1250-NEXT: s_and_b32 s6, s6, exec_lo -; GFX1250-NEXT: s_cselect_b32 s4, s4, s5 -; GFX1250-NEXT: v_mov_b32_e32 v2, s4 +; GFX1250-NEXT: v_cndmask_b32_e64 v0, 0, 1, s2 +; GFX1250-NEXT: s_and_b32 s2, s2, exec_lo +; GFX1250-NEXT: s_cselect_b32 s0, s0, s1 +; GFX1250-NEXT: v_mov_b32_e32 v2, s0 ; GFX1250-NEXT: s_clause 0x1 -; GFX1250-NEXT: global_store_b32 v1, v2, s[0:1] -; GFX1250-NEXT: global_store_b8 v1, v0, s[2:3] +; GFX1250-NEXT: global_store_b32 v1, v2, s[8:9] +; GFX1250-NEXT: global_store_b8 v1, v0, s[10:11] ; GFX1250-NEXT: s_endpgm %a = load i32, ptr addrspace(1) %aptr, align 4 %b = load i32, ptr addrspace(1) %bptr, align 4 @@ -3220,12 +3219,12 @@ define amdgpu_kernel void @v_test_umin_ult_i16_multi_use(ptr addrspace(1) %out0, ; ; GFX1250-LABEL: v_test_umin_ult_i16_multi_use: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x0 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x0 ; GFX1250-NEXT: v_mov_b32_e32 v0, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 -; GFX1250-NEXT: global_load_u16 v1, v0, s[6:7] -; GFX1250-NEXT: global_load_u16 v2, v0, s[4:5] +; GFX1250-NEXT: global_load_u16 v1, v0, s[14:15] +; GFX1250-NEXT: global_load_u16 v2, v0, s[12:13] ; GFX1250-NEXT: s_wait_loadcnt 0x1 ; GFX1250-NEXT: v_and_b32_e32 v3, 0xffff, v1 ; GFX1250-NEXT: s_wait_loadcnt 0x0 @@ -3235,8 +3234,8 @@ define amdgpu_kernel void @v_test_umin_ult_i16_multi_use(ptr addrspace(1) %out0, ; GFX1250-NEXT: v_cndmask_b32_e32 v1, v1, v2, vcc_lo ; GFX1250-NEXT: v_cndmask_b32_e64 v2, 0, 1, vcc_lo ; GFX1250-NEXT: s_clause 0x1 -; GFX1250-NEXT: global_store_b16 v0, v1, s[0:1] -; GFX1250-NEXT: global_store_b8 v0, v2, s[2:3] +; GFX1250-NEXT: global_store_b16 v0, v1, s[8:9] +; GFX1250-NEXT: global_store_b8 v0, v2, s[10:11] ; GFX1250-NEXT: s_endpgm %a = load i16, ptr addrspace(1) %aptr, align 2 %b = load i16, ptr addrspace(1) %bptr, align 2 @@ -4338,12 +4337,12 @@ define amdgpu_kernel void @test_umin_ult_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_umin_ult_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_min_u64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_min_u64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm %tmp = icmp ult i64 %a, %b @@ -4462,12 +4461,12 @@ define amdgpu_kernel void @test_umin_ule_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_umin_ule_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_min_u64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_min_u64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm %tmp = icmp ule i64 %a, %b @@ -4586,12 +4585,12 @@ define amdgpu_kernel void @test_imin_slt_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_imin_slt_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_min_i64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_min_i64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm %tmp = icmp slt i64 %a, %b @@ -4710,12 +4709,12 @@ define amdgpu_kernel void @test_imin_sle_i64(ptr addrspace(1) %out, i64 %a, i64 ; ; GFX1250-LABEL: test_imin_sle_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_min_i64 v[0:1], s[2:3], s[4:5] +; GFX1250-NEXT: v_min_i64 v[0:1], s[2:3], s[6:7] ; GFX1250-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX1250-NEXT: s_endpgm %tmp = icmp sle i64 %a, %b @@ -4872,14 +4871,14 @@ define amdgpu_kernel void @v_test_imin_sle_v2i16(ptr addrspace(1) %out, ptr addr ; ; GFX1250-LABEL: v_test_imin_sle_v2i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b32 v1, v0, s[2:3] scale_offset -; GFX1250-NEXT: global_load_b32 v2, v0, s[4:5] scale_offset +; GFX1250-NEXT: global_load_b32 v2, v0, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_pk_min_i16 v1, v1, v2 ; GFX1250-NEXT: global_store_b32 v0, v1, s[0:1] scale_offset @@ -5042,14 +5041,14 @@ define amdgpu_kernel void @v_test_imin_ule_v2i16(ptr addrspace(1) %out, ptr addr ; ; GFX1250-LABEL: v_test_imin_ule_v2i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: global_load_b32 v1, v0, s[2:3] scale_offset -; GFX1250-NEXT: global_load_b32 v2, v0, s[4:5] scale_offset +; GFX1250-NEXT: global_load_b32 v2, v0, s[6:7] scale_offset ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_pk_min_u16 v1, v1, v2 ; GFX1250-NEXT: global_store_b32 v0, v1, s[0:1] scale_offset diff --git a/llvm/test/CodeGen/AMDGPU/mul.ll b/llvm/test/CodeGen/AMDGPU/mul.ll index baccb4c..d29847e 100644 --- a/llvm/test/CodeGen/AMDGPU/mul.ll +++ b/llvm/test/CodeGen/AMDGPU/mul.ll @@ -450,6 +450,7 @@ define amdgpu_kernel void @s_trunc_i64_mul_to_i32(ptr addrspace(1) %out, i64 %a, ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_load_b32 s3, s[4:5], 0x34 +; GFX1250-NEXT: ; kill: killed $sgpr4_sgpr5 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_mul_i32 s2, s3, s2 ; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 @@ -613,25 +614,25 @@ define amdgpu_kernel void @v_trunc_i64_mul_to_i32(ptr addrspace(1) %out, ptr add ; ; GFX1250-LABEL: v_trunc_i64_mul_to_i32: ; GFX1250: ; %bb.0: ; %entry +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s6, s10 -; GFX1250-NEXT: s_mov_b32 s7, s11 +; GFX1250-NEXT: s_load_b64 s[8:9], s[4:5], 0x34 +; GFX1250-NEXT: s_mov_b32 s6, -1 +; GFX1250-NEXT: s_mov_b32 s7, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s14, s6 +; GFX1250-NEXT: s_mov_b32 s15, s7 +; GFX1250-NEXT: s_mov_b32 s10, s6 +; GFX1250-NEXT: s_mov_b32 s11, s7 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_mov_b32 s12, s2 ; GFX1250-NEXT: s_mov_b32 s13, s3 ; GFX1250-NEXT: buffer_load_b32 v0, off, s[12:15], null -; GFX1250-NEXT: buffer_load_b32 v1, off, s[4:7], null -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: buffer_load_b32 v1, off, s[8:11], null +; GFX1250-NEXT: s_mov_b32 s4, s0 +; GFX1250-NEXT: s_mov_b32 s5, s1 ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_mul_lo_u32 v0, v1, v0 -; GFX1250-NEXT: buffer_store_b32 v0, off, s[8:11], null +; GFX1250-NEXT: buffer_store_b32 v0, off, s[4:7], null ; GFX1250-NEXT: s_endpgm ; ; EG-LABEL: v_trunc_i64_mul_to_i32: @@ -2091,11 +2092,11 @@ define amdgpu_kernel void @s_mul_i64(ptr addrspace(1) %out, i64 %a, i64 %b) noun ; ; GFX1250-LABEL: s_mul_i64: ; GFX1250: ; %bb.0: ; %entry +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_mul_u64 s[4:5], s[2:3], s[4:5] +; GFX1250-NEXT: s_mul_u64 s[4:5], s[2:3], s[6:7] ; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 ; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[4:5] ; GFX1250-NEXT: s_mov_b32 s2, -1 @@ -2292,25 +2293,25 @@ define amdgpu_kernel void @v_mul_i64(ptr addrspace(1) %out, ptr addrspace(1) %ap ; ; GFX1250-LABEL: v_mul_i64: ; GFX1250: ; %bb.0: ; %entry +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 -; GFX1250-NEXT: s_mov_b32 s10, -1 -; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s14, s10 -; GFX1250-NEXT: s_mov_b32 s15, s11 -; GFX1250-NEXT: s_mov_b32 s6, s10 -; GFX1250-NEXT: s_mov_b32 s7, s11 +; GFX1250-NEXT: s_load_b64 s[8:9], s[4:5], 0x34 +; GFX1250-NEXT: s_mov_b32 s6, -1 +; GFX1250-NEXT: s_mov_b32 s7, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s14, s6 +; GFX1250-NEXT: s_mov_b32 s15, s7 +; GFX1250-NEXT: s_mov_b32 s10, s6 +; GFX1250-NEXT: s_mov_b32 s11, s7 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_mov_b32 s12, s2 ; GFX1250-NEXT: s_mov_b32 s13, s3 ; GFX1250-NEXT: buffer_load_b64 v[0:1], off, s[12:15], null -; GFX1250-NEXT: buffer_load_b64 v[2:3], off, s[4:7], null -; GFX1250-NEXT: s_mov_b32 s8, s0 -; GFX1250-NEXT: s_mov_b32 s9, s1 +; GFX1250-NEXT: buffer_load_b64 v[2:3], off, s[8:11], null +; GFX1250-NEXT: s_mov_b32 s4, s0 +; GFX1250-NEXT: s_mov_b32 s5, s1 ; GFX1250-NEXT: s_wait_loadcnt 0x0 ; GFX1250-NEXT: v_mul_u64_e32 v[0:1], v[0:1], v[2:3] -; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[8:11], null +; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[4:7], null ; GFX1250-NEXT: s_endpgm ; ; EG-LABEL: v_mul_i64: @@ -2845,30 +2846,30 @@ define amdgpu_kernel void @mul64_in_branch(ptr addrspace(1) %out, ptr addrspace( ; ; GFX1250-LABEL: mul64_in_branch: ; GFX1250: ; %bb.0: ; %entry -; GFX1250-NEXT: s_load_b256 s[0:7], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: s_cmp_lg_u64 s[4:5], 0 +; GFX1250-NEXT: s_cmp_lg_u64 s[12:13], 0 ; GFX1250-NEXT: s_cbranch_scc0 .LBB16_3 ; GFX1250-NEXT: ; %bb.1: ; %else -; GFX1250-NEXT: s_mul_u64 s[4:5], s[4:5], s[6:7] +; GFX1250-NEXT: s_mul_u64 s[0:1], s[12:13], s[14:15] ; GFX1250-NEXT: s_cbranch_execnz .LBB16_4 ; GFX1250-NEXT: .LBB16_2: ; %if -; GFX1250-NEXT: s_mov_b32 s7, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s6, -1 -; GFX1250-NEXT: s_mov_b32 s4, s2 -; GFX1250-NEXT: s_mov_b32 s5, s3 -; GFX1250-NEXT: buffer_load_b64 v[0:1], off, s[4:7], null +; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s0, s10 +; GFX1250-NEXT: s_mov_b32 s1, s11 +; GFX1250-NEXT: buffer_load_b64 v[0:1], off, s[0:3], null ; GFX1250-NEXT: s_branch .LBB16_5 ; GFX1250-NEXT: .LBB16_3: -; GFX1250-NEXT: ; implicit-def: $sgpr4_sgpr5 +; GFX1250-NEXT: ; implicit-def: $sgpr0_sgpr1 ; GFX1250-NEXT: s_branch .LBB16_2 ; GFX1250-NEXT: .LBB16_4: -; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[4:5] +; GFX1250-NEXT: v_mov_b64_e32 v[0:1], s[0:1] ; GFX1250-NEXT: .LBB16_5: ; %endif -; GFX1250-NEXT: s_mov_b32 s3, 0x31016000 -; GFX1250-NEXT: s_mov_b32 s2, -1 +; GFX1250-NEXT: s_mov_b32 s11, 0x31016000 +; GFX1250-NEXT: s_mov_b32 s10, -1 ; GFX1250-NEXT: s_wait_loadcnt 0x0 -; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[0:3], null +; GFX1250-NEXT: buffer_store_b64 v[0:1], off, s[8:11], null ; GFX1250-NEXT: s_endpgm ; ; EG-LABEL: mul64_in_branch: diff --git a/llvm/test/CodeGen/AMDGPU/packed-fp32.ll b/llvm/test/CodeGen/AMDGPU/packed-fp32.ll index b0651ef..78207c2 100644 --- a/llvm/test/CodeGen/AMDGPU/packed-fp32.ll +++ b/llvm/test/CodeGen/AMDGPU/packed-fp32.ll @@ -340,46 +340,46 @@ define amdgpu_kernel void @fadd_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; ; GFX1250-SDAG-LABEL: fadd_v32_vs: ; GFX1250-SDAG: ; %bb.0: -; GFX1250-SDAG-NEXT: s_load_b64 s[34:35], s[4:5], 0x24 +; GFX1250-SDAG-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 ; GFX1250-SDAG-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_lshlrev_b32_e32 v56, 7, v0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_clause 0x7 -; GFX1250-SDAG-NEXT: global_load_b128 v[0:3], v56, s[34:35] offset:16 -; GFX1250-SDAG-NEXT: global_load_b128 v[4:7], v56, s[34:35] offset:48 -; GFX1250-SDAG-NEXT: global_load_b128 v[8:11], v56, s[34:35] offset:32 -; GFX1250-SDAG-NEXT: global_load_b128 v[12:15], v56, s[34:35] -; GFX1250-SDAG-NEXT: global_load_b128 v[16:19], v56, s[34:35] offset:80 -; GFX1250-SDAG-NEXT: global_load_b128 v[20:23], v56, s[34:35] offset:96 -; GFX1250-SDAG-NEXT: global_load_b128 v[24:27], v56, s[34:35] offset:64 -; GFX1250-SDAG-NEXT: global_load_b128 v[28:31], v56, s[34:35] offset:112 -; GFX1250-SDAG-NEXT: s_load_b512 s[16:31], s[4:5], 0xa4 -; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 -; GFX1250-SDAG-NEXT: s_load_b512 s[0:15], s[4:5], 0xe4 +; GFX1250-SDAG-NEXT: global_load_b128 v[0:3], v56, s[0:1] offset:16 +; GFX1250-SDAG-NEXT: global_load_b128 v[4:7], v56, s[0:1] offset:48 +; GFX1250-SDAG-NEXT: global_load_b128 v[8:11], v56, s[0:1] offset:32 +; GFX1250-SDAG-NEXT: global_load_b128 v[12:15], v56, s[0:1] +; GFX1250-SDAG-NEXT: global_load_b128 v[16:19], v56, s[0:1] offset:80 +; GFX1250-SDAG-NEXT: global_load_b128 v[20:23], v56, s[0:1] offset:96 +; GFX1250-SDAG-NEXT: global_load_b128 v[24:27], v56, s[0:1] offset:64 +; GFX1250-SDAG-NEXT: global_load_b128 v[28:31], v56, s[0:1] offset:112 +; GFX1250-SDAG-NEXT: s_clause 0x1 +; GFX1250-SDAG-NEXT: s_load_b512 s[36:51], s[4:5], 0xa4 +; GFX1250-SDAG-NEXT: s_load_b512 s[8:23], s[4:5], 0xe4 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s20 :: v_dual_mov_b32 v33, s21 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s22 :: v_dual_mov_b32 v35, s23 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v36, s18 :: v_dual_mov_b32 v39, s29 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v40, s30 :: v_dual_mov_b32 v41, s31 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v42, s24 :: v_dual_mov_b32 v37, s19 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v38, s28 :: v_dual_mov_b32 v55, s15 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v51, s3 :: v_dual_mov_b32 v52, s12 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v53, s13 :: v_dual_mov_b32 v54, s14 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v49, s7 :: v_dual_mov_b32 v50, s2 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v45, s27 :: v_dual_mov_b32 v46, s4 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v47, s5 :: v_dual_mov_b32 v48, s6 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v43, s25 :: v_dual_mov_b32 v44, s26 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s40 :: v_dual_mov_b32 v33, s41 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s42 :: v_dual_mov_b32 v35, s43 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v36, s38 :: v_dual_mov_b32 v39, s49 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v40, s50 :: v_dual_mov_b32 v41, s51 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v42, s44 :: v_dual_mov_b32 v37, s39 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v38, s48 :: v_dual_mov_b32 v55, s23 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v51, s11 :: v_dual_mov_b32 v52, s20 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v53, s21 :: v_dual_mov_b32 v54, s22 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v49, s15 :: v_dual_mov_b32 v50, s10 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v45, s47 :: v_dual_mov_b32 v46, s12 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v47, s13 :: v_dual_mov_b32 v48, s14 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v43, s45 :: v_dual_mov_b32 v44, s46 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x7 ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[0:1], v[0:1], v[32:33] ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[2:3], v[2:3], v[34:35] -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s8 :: v_dual_mov_b32 v33, s9 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s10 :: v_dual_mov_b32 v35, s11 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s16 :: v_dual_mov_b32 v33, s17 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s18 :: v_dual_mov_b32 v35, s19 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x6 ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[6:7], v[6:7], v[40:41] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[0:1] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[8:9] ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[4:5], v[4:5], v[38:39] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[16:17] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[36:37] ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x2 ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[20:21], v[20:21], v[32:33] ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[22:23], v[22:23], v[34:35] @@ -395,58 +395,58 @@ define amdgpu_kernel void @fadd_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[14:15], v[14:15], v[36:37] ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[12:13], v[12:13], v[38:39] ; GFX1250-SDAG-NEXT: s_clause 0x7 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[20:23], s[34:35] offset:96 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[28:31], s[34:35] offset:112 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[24:27], s[34:35] offset:64 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[16:19], s[34:35] offset:80 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[8:11], s[34:35] offset:32 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[4:7], s[34:35] offset:48 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[12:15], s[34:35] -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[0:3], s[34:35] offset:16 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[20:23], s[0:1] offset:96 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[28:31], s[0:1] offset:112 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[24:27], s[0:1] offset:64 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[16:19], s[0:1] offset:80 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[8:11], s[0:1] offset:32 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[4:7], s[0:1] offset:48 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[12:15], s[0:1] +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[0:3], s[0:1] offset:16 ; GFX1250-SDAG-NEXT: s_endpgm ; ; GFX1250-GISEL-LABEL: fadd_v32_vs: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: s_load_b64 s[34:35], s[4:5], 0x24 +; GFX1250-GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 ; GFX1250-GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_lshlrev_b32_e32 v56, 7, v0 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_clause 0x7 -; GFX1250-GISEL-NEXT: global_load_b128 v[0:3], v56, s[34:35] -; GFX1250-GISEL-NEXT: global_load_b128 v[4:7], v56, s[34:35] offset:16 -; GFX1250-GISEL-NEXT: global_load_b128 v[8:11], v56, s[34:35] offset:32 -; GFX1250-GISEL-NEXT: global_load_b128 v[12:15], v56, s[34:35] offset:48 -; GFX1250-GISEL-NEXT: global_load_b128 v[16:19], v56, s[34:35] offset:64 -; GFX1250-GISEL-NEXT: global_load_b128 v[20:23], v56, s[34:35] offset:80 -; GFX1250-GISEL-NEXT: global_load_b128 v[24:27], v56, s[34:35] offset:96 -; GFX1250-GISEL-NEXT: global_load_b128 v[28:31], v56, s[34:35] offset:112 -; GFX1250-GISEL-NEXT: s_load_b512 s[16:31], s[4:5], 0xa4 -; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 -; GFX1250-GISEL-NEXT: s_load_b512 s[0:15], s[4:5], 0xe4 +; GFX1250-GISEL-NEXT: global_load_b128 v[0:3], v56, s[0:1] +; GFX1250-GISEL-NEXT: global_load_b128 v[4:7], v56, s[0:1] offset:16 +; GFX1250-GISEL-NEXT: global_load_b128 v[8:11], v56, s[0:1] offset:32 +; GFX1250-GISEL-NEXT: global_load_b128 v[12:15], v56, s[0:1] offset:48 +; GFX1250-GISEL-NEXT: global_load_b128 v[16:19], v56, s[0:1] offset:64 +; GFX1250-GISEL-NEXT: global_load_b128 v[20:23], v56, s[0:1] offset:80 +; GFX1250-GISEL-NEXT: global_load_b128 v[24:27], v56, s[0:1] offset:96 +; GFX1250-GISEL-NEXT: global_load_b128 v[28:31], v56, s[0:1] offset:112 +; GFX1250-GISEL-NEXT: s_clause 0x1 +; GFX1250-GISEL-NEXT: s_load_b512 s[36:51], s[4:5], 0xa4 +; GFX1250-GISEL-NEXT: s_load_b512 s[8:23], s[4:5], 0xe4 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[16:17] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[18:19] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[20:21] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[22:23] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[40:41], s[24:25] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[42:43], s[26:27] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[44:45], s[28:29] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[46:47], s[30:31] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[48:49], s[0:1] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[50:51], s[2:3] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[52:53], s[4:5] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[54:55], s[6:7] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[36:37] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[38:39] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[40:41] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[42:43] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[40:41], s[44:45] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[42:43], s[46:47] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[44:45], s[48:49] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[46:47], s[50:51] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[48:49], s[8:9] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[50:51], s[10:11] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[52:53], s[12:13] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[54:55], s[14:15] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x7 ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[0:1], v[0:1], v[32:33] ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[2:3], v[2:3], v[34:35] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[8:9] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[10:11] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[16:17] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[18:19] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x6 ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[4:5], v[4:5], v[36:37] ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[6:7], v[6:7], v[38:39] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[12:13] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[14:15] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[20:21] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[22:23] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x5 ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[8:9], v[8:9], v[40:41] ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[10:11], v[10:11], v[42:43] @@ -466,14 +466,14 @@ define amdgpu_kernel void @fadd_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[28:29], v[28:29], v[36:37] ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[30:31], v[30:31], v[38:39] ; GFX1250-GISEL-NEXT: s_clause 0x7 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[0:3], s[34:35] -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[4:7], s[34:35] offset:16 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[8:11], s[34:35] offset:32 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[12:15], s[34:35] offset:48 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[16:19], s[34:35] offset:64 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[20:23], s[34:35] offset:80 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[24:27], s[34:35] offset:96 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[28:31], s[34:35] offset:112 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[0:3], s[0:1] +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[4:7], s[0:1] offset:16 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[8:11], s[0:1] offset:32 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[12:15], s[0:1] offset:48 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[16:19], s[0:1] offset:64 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[20:23], s[0:1] offset:80 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[24:27], s[0:1] offset:96 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[28:31], s[0:1] offset:112 ; GFX1250-GISEL-NEXT: s_endpgm %id = tail call i32 @llvm.amdgcn.workitem.id.x() %gep = getelementptr inbounds <32 x float>, ptr addrspace(1) %a, i32 %id @@ -1597,46 +1597,46 @@ define amdgpu_kernel void @fmul_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; ; GFX1250-SDAG-LABEL: fmul_v32_vs: ; GFX1250-SDAG: ; %bb.0: -; GFX1250-SDAG-NEXT: s_load_b64 s[34:35], s[4:5], 0x24 +; GFX1250-SDAG-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 ; GFX1250-SDAG-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_lshlrev_b32_e32 v56, 7, v0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_clause 0x7 -; GFX1250-SDAG-NEXT: global_load_b128 v[0:3], v56, s[34:35] offset:16 -; GFX1250-SDAG-NEXT: global_load_b128 v[4:7], v56, s[34:35] offset:48 -; GFX1250-SDAG-NEXT: global_load_b128 v[8:11], v56, s[34:35] offset:32 -; GFX1250-SDAG-NEXT: global_load_b128 v[12:15], v56, s[34:35] -; GFX1250-SDAG-NEXT: global_load_b128 v[16:19], v56, s[34:35] offset:80 -; GFX1250-SDAG-NEXT: global_load_b128 v[20:23], v56, s[34:35] offset:96 -; GFX1250-SDAG-NEXT: global_load_b128 v[24:27], v56, s[34:35] offset:64 -; GFX1250-SDAG-NEXT: global_load_b128 v[28:31], v56, s[34:35] offset:112 -; GFX1250-SDAG-NEXT: s_load_b512 s[16:31], s[4:5], 0xa4 -; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 -; GFX1250-SDAG-NEXT: s_load_b512 s[0:15], s[4:5], 0xe4 +; GFX1250-SDAG-NEXT: global_load_b128 v[0:3], v56, s[0:1] offset:16 +; GFX1250-SDAG-NEXT: global_load_b128 v[4:7], v56, s[0:1] offset:48 +; GFX1250-SDAG-NEXT: global_load_b128 v[8:11], v56, s[0:1] offset:32 +; GFX1250-SDAG-NEXT: global_load_b128 v[12:15], v56, s[0:1] +; GFX1250-SDAG-NEXT: global_load_b128 v[16:19], v56, s[0:1] offset:80 +; GFX1250-SDAG-NEXT: global_load_b128 v[20:23], v56, s[0:1] offset:96 +; GFX1250-SDAG-NEXT: global_load_b128 v[24:27], v56, s[0:1] offset:64 +; GFX1250-SDAG-NEXT: global_load_b128 v[28:31], v56, s[0:1] offset:112 +; GFX1250-SDAG-NEXT: s_clause 0x1 +; GFX1250-SDAG-NEXT: s_load_b512 s[36:51], s[4:5], 0xa4 +; GFX1250-SDAG-NEXT: s_load_b512 s[8:23], s[4:5], 0xe4 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s20 :: v_dual_mov_b32 v33, s21 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s22 :: v_dual_mov_b32 v35, s23 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v36, s18 :: v_dual_mov_b32 v39, s29 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v40, s30 :: v_dual_mov_b32 v41, s31 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v42, s24 :: v_dual_mov_b32 v37, s19 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v38, s28 :: v_dual_mov_b32 v55, s15 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v51, s3 :: v_dual_mov_b32 v52, s12 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v53, s13 :: v_dual_mov_b32 v54, s14 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v49, s7 :: v_dual_mov_b32 v50, s2 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v45, s27 :: v_dual_mov_b32 v46, s4 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v47, s5 :: v_dual_mov_b32 v48, s6 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v43, s25 :: v_dual_mov_b32 v44, s26 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s40 :: v_dual_mov_b32 v33, s41 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s42 :: v_dual_mov_b32 v35, s43 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v36, s38 :: v_dual_mov_b32 v39, s49 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v40, s50 :: v_dual_mov_b32 v41, s51 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v42, s44 :: v_dual_mov_b32 v37, s39 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v38, s48 :: v_dual_mov_b32 v55, s23 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v51, s11 :: v_dual_mov_b32 v52, s20 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v53, s21 :: v_dual_mov_b32 v54, s22 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v49, s15 :: v_dual_mov_b32 v50, s10 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v45, s47 :: v_dual_mov_b32 v46, s12 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v47, s13 :: v_dual_mov_b32 v48, s14 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v43, s45 :: v_dual_mov_b32 v44, s46 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x7 ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[0:1], v[0:1], v[32:33] ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[2:3], v[2:3], v[34:35] -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s8 :: v_dual_mov_b32 v33, s9 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s10 :: v_dual_mov_b32 v35, s11 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v32, s16 :: v_dual_mov_b32 v33, s17 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v34, s18 :: v_dual_mov_b32 v35, s19 ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x6 ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[6:7], v[6:7], v[40:41] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[0:1] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[8:9] ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[4:5], v[4:5], v[38:39] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[16:17] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[36:37] ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x2 ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[20:21], v[20:21], v[32:33] ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[22:23], v[22:23], v[34:35] @@ -1652,58 +1652,58 @@ define amdgpu_kernel void @fmul_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[14:15], v[14:15], v[36:37] ; GFX1250-SDAG-NEXT: v_pk_mul_f32 v[12:13], v[12:13], v[38:39] ; GFX1250-SDAG-NEXT: s_clause 0x7 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[20:23], s[34:35] offset:96 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[28:31], s[34:35] offset:112 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[24:27], s[34:35] offset:64 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[16:19], s[34:35] offset:80 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[8:11], s[34:35] offset:32 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[4:7], s[34:35] offset:48 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[12:15], s[34:35] -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[0:3], s[34:35] offset:16 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[20:23], s[0:1] offset:96 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[28:31], s[0:1] offset:112 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[24:27], s[0:1] offset:64 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[16:19], s[0:1] offset:80 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[8:11], s[0:1] offset:32 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[4:7], s[0:1] offset:48 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[12:15], s[0:1] +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[0:3], s[0:1] offset:16 ; GFX1250-SDAG-NEXT: s_endpgm ; ; GFX1250-GISEL-LABEL: fmul_v32_vs: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: s_load_b64 s[34:35], s[4:5], 0x24 +; GFX1250-GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 ; GFX1250-GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_lshlrev_b32_e32 v56, 7, v0 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_clause 0x7 -; GFX1250-GISEL-NEXT: global_load_b128 v[0:3], v56, s[34:35] -; GFX1250-GISEL-NEXT: global_load_b128 v[4:7], v56, s[34:35] offset:16 -; GFX1250-GISEL-NEXT: global_load_b128 v[8:11], v56, s[34:35] offset:32 -; GFX1250-GISEL-NEXT: global_load_b128 v[12:15], v56, s[34:35] offset:48 -; GFX1250-GISEL-NEXT: global_load_b128 v[16:19], v56, s[34:35] offset:64 -; GFX1250-GISEL-NEXT: global_load_b128 v[20:23], v56, s[34:35] offset:80 -; GFX1250-GISEL-NEXT: global_load_b128 v[24:27], v56, s[34:35] offset:96 -; GFX1250-GISEL-NEXT: global_load_b128 v[28:31], v56, s[34:35] offset:112 -; GFX1250-GISEL-NEXT: s_load_b512 s[16:31], s[4:5], 0xa4 -; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 -; GFX1250-GISEL-NEXT: s_load_b512 s[0:15], s[4:5], 0xe4 +; GFX1250-GISEL-NEXT: global_load_b128 v[0:3], v56, s[0:1] +; GFX1250-GISEL-NEXT: global_load_b128 v[4:7], v56, s[0:1] offset:16 +; GFX1250-GISEL-NEXT: global_load_b128 v[8:11], v56, s[0:1] offset:32 +; GFX1250-GISEL-NEXT: global_load_b128 v[12:15], v56, s[0:1] offset:48 +; GFX1250-GISEL-NEXT: global_load_b128 v[16:19], v56, s[0:1] offset:64 +; GFX1250-GISEL-NEXT: global_load_b128 v[20:23], v56, s[0:1] offset:80 +; GFX1250-GISEL-NEXT: global_load_b128 v[24:27], v56, s[0:1] offset:96 +; GFX1250-GISEL-NEXT: global_load_b128 v[28:31], v56, s[0:1] offset:112 +; GFX1250-GISEL-NEXT: s_clause 0x1 +; GFX1250-GISEL-NEXT: s_load_b512 s[36:51], s[4:5], 0xa4 +; GFX1250-GISEL-NEXT: s_load_b512 s[8:23], s[4:5], 0xe4 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[16:17] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[18:19] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[20:21] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[22:23] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[40:41], s[24:25] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[42:43], s[26:27] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[44:45], s[28:29] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[46:47], s[30:31] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[48:49], s[0:1] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[50:51], s[2:3] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[52:53], s[4:5] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[54:55], s[6:7] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[36:37] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[38:39] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[40:41] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[42:43] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[40:41], s[44:45] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[42:43], s[46:47] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[44:45], s[48:49] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[46:47], s[50:51] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[48:49], s[8:9] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[50:51], s[10:11] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[52:53], s[12:13] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[54:55], s[14:15] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x7 ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[0:1], v[0:1], v[32:33] ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[2:3], v[2:3], v[34:35] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[8:9] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[10:11] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[16:17] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[18:19] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x6 ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[4:5], v[4:5], v[36:37] ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[6:7], v[6:7], v[38:39] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[12:13] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[14:15] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[20:21] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[22:23] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x5 ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[8:9], v[8:9], v[40:41] ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[10:11], v[10:11], v[42:43] @@ -1723,14 +1723,14 @@ define amdgpu_kernel void @fmul_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[28:29], v[28:29], v[36:37] ; GFX1250-GISEL-NEXT: v_pk_mul_f32 v[30:31], v[30:31], v[38:39] ; GFX1250-GISEL-NEXT: s_clause 0x7 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[0:3], s[34:35] -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[4:7], s[34:35] offset:16 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[8:11], s[34:35] offset:32 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[12:15], s[34:35] offset:48 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[16:19], s[34:35] offset:64 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[20:23], s[34:35] offset:80 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[24:27], s[34:35] offset:96 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[28:31], s[34:35] offset:112 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[0:3], s[0:1] +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[4:7], s[0:1] offset:16 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[8:11], s[0:1] offset:32 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[12:15], s[0:1] offset:48 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[16:19], s[0:1] offset:64 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[20:23], s[0:1] offset:80 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[24:27], s[0:1] offset:96 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[28:31], s[0:1] offset:112 ; GFX1250-GISEL-NEXT: s_endpgm %id = tail call i32 @llvm.amdgcn.workitem.id.x() %gep = getelementptr inbounds <32 x float>, ptr addrspace(1) %a, i32 %id @@ -2428,46 +2428,46 @@ define amdgpu_kernel void @fma_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; ; GFX1250-SDAG-LABEL: fma_v32_vs: ; GFX1250-SDAG: ; %bb.0: -; GFX1250-SDAG-NEXT: s_load_b64 s[34:35], s[4:5], 0x24 +; GFX1250-SDAG-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 ; GFX1250-SDAG-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-SDAG-NEXT: v_lshlrev_b32_e32 v56, 7, v0 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_clause 0x7 -; GFX1250-SDAG-NEXT: global_load_b128 v[0:3], v56, s[34:35] offset:16 -; GFX1250-SDAG-NEXT: global_load_b128 v[4:7], v56, s[34:35] offset:48 -; GFX1250-SDAG-NEXT: global_load_b128 v[8:11], v56, s[34:35] offset:32 -; GFX1250-SDAG-NEXT: global_load_b128 v[12:15], v56, s[34:35] -; GFX1250-SDAG-NEXT: global_load_b128 v[16:19], v56, s[34:35] offset:80 -; GFX1250-SDAG-NEXT: global_load_b128 v[20:23], v56, s[34:35] offset:96 -; GFX1250-SDAG-NEXT: global_load_b128 v[24:27], v56, s[34:35] offset:64 -; GFX1250-SDAG-NEXT: global_load_b128 v[28:31], v56, s[34:35] offset:112 -; GFX1250-SDAG-NEXT: s_load_b512 s[16:31], s[4:5], 0xa4 -; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 -; GFX1250-SDAG-NEXT: s_load_b512 s[0:15], s[4:5], 0xe4 +; GFX1250-SDAG-NEXT: global_load_b128 v[0:3], v56, s[0:1] offset:16 +; GFX1250-SDAG-NEXT: global_load_b128 v[4:7], v56, s[0:1] offset:48 +; GFX1250-SDAG-NEXT: global_load_b128 v[8:11], v56, s[0:1] offset:32 +; GFX1250-SDAG-NEXT: global_load_b128 v[12:15], v56, s[0:1] +; GFX1250-SDAG-NEXT: global_load_b128 v[16:19], v56, s[0:1] offset:80 +; GFX1250-SDAG-NEXT: global_load_b128 v[20:23], v56, s[0:1] offset:96 +; GFX1250-SDAG-NEXT: global_load_b128 v[24:27], v56, s[0:1] offset:64 +; GFX1250-SDAG-NEXT: global_load_b128 v[28:31], v56, s[0:1] offset:112 +; GFX1250-SDAG-NEXT: s_clause 0x1 +; GFX1250-SDAG-NEXT: s_load_b512 s[36:51], s[4:5], 0xa4 +; GFX1250-SDAG-NEXT: s_load_b512 s[8:23], s[4:5], 0xe4 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[32:33], s[20:21] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[34:35], s[22:23] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[30:31] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[28:29] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[52:53], s[12:13] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[54:55], s[14:15] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[50:51], s[2:3] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[46:47], s[4:5] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[48:49], s[6:7] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[42:43], s[24:25] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[44:45], s[26:27] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[36:37], s[18:19] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[32:33], s[40:41] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[34:35], s[42:43] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[50:51] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[48:49] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[52:53], s[20:21] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[54:55], s[22:23] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[50:51], s[10:11] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[46:47], s[12:13] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[48:49], s[14:15] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[42:43], s[44:45] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[44:45], s[46:47] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[36:37], s[38:39] ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x7 ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[0:1], v[0:1], v[32:33], v[32:33] ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[2:3], v[2:3], v[34:35], v[34:35] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[32:33], s[8:9] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[34:35], s[10:11] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[32:33], s[16:17] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[34:35], s[18:19] ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x6 ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[6:7], v[6:7], v[40:41], v[40:41] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[0:1] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[40:41], s[8:9] ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[4:5], v[4:5], v[38:39], v[38:39] -; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[16:17] +; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[38:39], s[36:37] ; GFX1250-SDAG-NEXT: s_wait_loadcnt 0x0 ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[28:29], v[28:29], v[52:53], v[52:53] ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[20:21], v[20:21], v[32:33], v[32:33] @@ -2482,58 +2482,58 @@ define amdgpu_kernel void @fma_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[14:15], v[14:15], v[36:37], v[36:37] ; GFX1250-SDAG-NEXT: v_pk_fma_f32 v[12:13], v[12:13], v[38:39], v[38:39] ; GFX1250-SDAG-NEXT: s_clause 0x7 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[20:23], s[34:35] offset:96 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[28:31], s[34:35] offset:112 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[24:27], s[34:35] offset:64 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[16:19], s[34:35] offset:80 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[8:11], s[34:35] offset:32 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[4:7], s[34:35] offset:48 -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[12:15], s[34:35] -; GFX1250-SDAG-NEXT: global_store_b128 v56, v[0:3], s[34:35] offset:16 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[20:23], s[0:1] offset:96 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[28:31], s[0:1] offset:112 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[24:27], s[0:1] offset:64 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[16:19], s[0:1] offset:80 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[8:11], s[0:1] offset:32 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[4:7], s[0:1] offset:48 +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[12:15], s[0:1] +; GFX1250-SDAG-NEXT: global_store_b128 v56, v[0:3], s[0:1] offset:16 ; GFX1250-SDAG-NEXT: s_endpgm ; ; GFX1250-GISEL-LABEL: fma_v32_vs: ; GFX1250-GISEL: ; %bb.0: -; GFX1250-GISEL-NEXT: s_load_b64 s[34:35], s[4:5], 0x24 +; GFX1250-GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 ; GFX1250-GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-GISEL-NEXT: v_lshlrev_b32_e32 v56, 7, v0 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_clause 0x7 -; GFX1250-GISEL-NEXT: global_load_b128 v[0:3], v56, s[34:35] -; GFX1250-GISEL-NEXT: global_load_b128 v[4:7], v56, s[34:35] offset:16 -; GFX1250-GISEL-NEXT: global_load_b128 v[8:11], v56, s[34:35] offset:32 -; GFX1250-GISEL-NEXT: global_load_b128 v[12:15], v56, s[34:35] offset:48 -; GFX1250-GISEL-NEXT: global_load_b128 v[16:19], v56, s[34:35] offset:64 -; GFX1250-GISEL-NEXT: global_load_b128 v[20:23], v56, s[34:35] offset:80 -; GFX1250-GISEL-NEXT: global_load_b128 v[24:27], v56, s[34:35] offset:96 -; GFX1250-GISEL-NEXT: global_load_b128 v[28:31], v56, s[34:35] offset:112 -; GFX1250-GISEL-NEXT: s_load_b512 s[16:31], s[4:5], 0xa4 -; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 -; GFX1250-GISEL-NEXT: s_load_b512 s[0:15], s[4:5], 0xe4 +; GFX1250-GISEL-NEXT: global_load_b128 v[0:3], v56, s[0:1] +; GFX1250-GISEL-NEXT: global_load_b128 v[4:7], v56, s[0:1] offset:16 +; GFX1250-GISEL-NEXT: global_load_b128 v[8:11], v56, s[0:1] offset:32 +; GFX1250-GISEL-NEXT: global_load_b128 v[12:15], v56, s[0:1] offset:48 +; GFX1250-GISEL-NEXT: global_load_b128 v[16:19], v56, s[0:1] offset:64 +; GFX1250-GISEL-NEXT: global_load_b128 v[20:23], v56, s[0:1] offset:80 +; GFX1250-GISEL-NEXT: global_load_b128 v[24:27], v56, s[0:1] offset:96 +; GFX1250-GISEL-NEXT: global_load_b128 v[28:31], v56, s[0:1] offset:112 +; GFX1250-GISEL-NEXT: s_clause 0x1 +; GFX1250-GISEL-NEXT: s_load_b512 s[36:51], s[4:5], 0xa4 +; GFX1250-GISEL-NEXT: s_load_b512 s[8:23], s[4:5], 0xe4 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[16:17] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[18:19] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[20:21] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[22:23] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[40:41], s[24:25] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[42:43], s[26:27] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[44:45], s[28:29] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[46:47], s[30:31] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[48:49], s[0:1] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[50:51], s[2:3] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[52:53], s[4:5] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[54:55], s[6:7] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[36:37] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[38:39] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[40:41] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[42:43] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[40:41], s[44:45] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[42:43], s[46:47] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[44:45], s[48:49] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[46:47], s[50:51] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[48:49], s[8:9] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[50:51], s[10:11] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[52:53], s[12:13] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[54:55], s[14:15] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x7 ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[0:1], v[0:1], v[32:33], v[32:33] ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[2:3], v[2:3], v[34:35], v[34:35] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[8:9] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[10:11] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[32:33], s[16:17] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[34:35], s[18:19] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x6 ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[4:5], v[4:5], v[36:37], v[36:37] ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[6:7], v[6:7], v[38:39], v[38:39] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[12:13] -; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[14:15] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[36:37], s[20:21] +; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[38:39], s[22:23] ; GFX1250-GISEL-NEXT: s_wait_loadcnt 0x5 ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[8:9], v[8:9], v[40:41], v[40:41] ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[10:11], v[10:11], v[42:43], v[42:43] @@ -2553,14 +2553,14 @@ define amdgpu_kernel void @fma_v32_vs(ptr addrspace(1) %a, <32 x float> %x) { ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[28:29], v[28:29], v[36:37], v[36:37] ; GFX1250-GISEL-NEXT: v_pk_fma_f32 v[30:31], v[30:31], v[38:39], v[38:39] ; GFX1250-GISEL-NEXT: s_clause 0x7 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[0:3], s[34:35] -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[4:7], s[34:35] offset:16 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[8:11], s[34:35] offset:32 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[12:15], s[34:35] offset:48 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[16:19], s[34:35] offset:64 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[20:23], s[34:35] offset:80 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[24:27], s[34:35] offset:96 -; GFX1250-GISEL-NEXT: global_store_b128 v56, v[28:31], s[34:35] offset:112 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[0:3], s[0:1] +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[4:7], s[0:1] offset:16 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[8:11], s[0:1] offset:32 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[12:15], s[0:1] offset:48 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[16:19], s[0:1] offset:64 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[20:23], s[0:1] offset:80 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[24:27], s[0:1] offset:96 +; GFX1250-GISEL-NEXT: global_store_b128 v56, v[28:31], s[0:1] offset:112 ; GFX1250-GISEL-NEXT: s_endpgm %id = tail call i32 @llvm.amdgcn.workitem.id.x() %gep = getelementptr inbounds <32 x float>, ptr addrspace(1) %a, i32 %id @@ -3529,9 +3529,9 @@ define amdgpu_kernel void @fadd_fadd_fsub(<2 x float> %arg, <2 x float> %arg1, p ; ; GFX1250-SDAG-LABEL: fadd_fadd_fsub: ; GFX1250-SDAG: ; %bb.0: ; %bb +; GFX1250-SDAG-NEXT: s_clause 0x1 ; GFX1250-SDAG-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 -; GFX1250-SDAG-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-SDAG-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX1250-SDAG-NEXT: s_add_f32 s2, s1, s3 @@ -3541,14 +3541,14 @@ define amdgpu_kernel void @fadd_fadd_fsub(<2 x float> %arg, <2 x float> %arg1, p ; GFX1250-SDAG-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, v2 ; GFX1250-SDAG-NEXT: v_mov_b32_e32 v2, 0 ; GFX1250-SDAG-NEXT: v_pk_add_f32 v[0:1], v[4:5], v[0:1] neg_lo:[0,1] neg_hi:[0,1] -; GFX1250-SDAG-NEXT: global_store_b64 v2, v[0:1], s[4:5] +; GFX1250-SDAG-NEXT: global_store_b64 v2, v[0:1], s[6:7] ; GFX1250-SDAG-NEXT: s_endpgm ; ; GFX1250-GISEL-LABEL: fadd_fadd_fsub: ; GFX1250-GISEL: ; %bb.0: ; %bb +; GFX1250-GISEL-NEXT: s_clause 0x1 ; GFX1250-GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 -; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 -; GFX1250-GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1250-GISEL-NEXT: s_load_b64 s[6:7], s[4:5], 0x34 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[0:1], s[0:1] ; GFX1250-GISEL-NEXT: v_mov_b64_e32 v[2:3], s[2:3] @@ -3560,7 +3560,7 @@ define amdgpu_kernel void @fadd_fadd_fsub(<2 x float> %arg, <2 x float> %arg1, p ; GFX1250-GISEL-NEXT: v_pk_add_f32 v[0:1], v[2:3], v[0:1] ; GFX1250-GISEL-NEXT: v_dual_mov_b32 v2, s0 :: v_dual_subrev_f32 v3, s3, v0 ; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, 0 -; GFX1250-GISEL-NEXT: global_store_b64 v0, v[2:3], s[4:5] +; GFX1250-GISEL-NEXT: global_store_b64 v0, v[2:3], s[6:7] ; GFX1250-GISEL-NEXT: s_endpgm bb: %i12 = fadd <2 x float> %arg, %arg1 diff --git a/llvm/test/CodeGen/AMDGPU/preload-implicit-kernargs.ll b/llvm/test/CodeGen/AMDGPU/preload-implicit-kernargs.ll index b717f85..6671201 100644 --- a/llvm/test/CodeGen/AMDGPU/preload-implicit-kernargs.ll +++ b/llvm/test/CodeGen/AMDGPU/preload-implicit-kernargs.ll @@ -186,12 +186,12 @@ define amdgpu_kernel void @mixed_inreg_block_count_x(ptr addrspace(1) %out, i32 ; ; GFX1250-LABEL: mixed_inreg_block_count_x: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b32 s2, s[0:1], 0x10 -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[0:1], s[0:1], 0x0 +; GFX1250-NEXT: s_clause 0x1 +; GFX1250-NEXT: s_load_b32 s4, s[0:1], 0x10 +; GFX1250-NEXT: s_load_b64 s[2:3], s[0:1], 0x0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s2 -; GFX1250-NEXT: global_store_b32 v0, v1, s[0:1] +; GFX1250-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s4 +; GFX1250-NEXT: global_store_b32 v0, v1, s[2:3] ; GFX1250-NEXT: s_endpgm %imp_arg_ptr = call ptr addrspace(4) @llvm.amdgcn.implicitarg.ptr() %load = load i32, ptr addrspace(4) %imp_arg_ptr diff --git a/llvm/test/CodeGen/AMDGPU/preload-kernargs.ll b/llvm/test/CodeGen/AMDGPU/preload-kernargs.ll index 4d367ef..c1764c9 100644 --- a/llvm/test/CodeGen/AMDGPU/preload-kernargs.ll +++ b/llvm/test/CodeGen/AMDGPU/preload-kernargs.ll @@ -346,10 +346,10 @@ define amdgpu_kernel void @byref_preload_arg(ptr addrspace(1) inreg %out, ptr ad ; ; GFX1250-LABEL: byref_preload_arg: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b64 s[0:1], s[0:1], 0x100 +; GFX1250-NEXT: s_load_b64 s[4:5], s[0:1], 0x100 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s0 -; GFX1250-NEXT: v_mov_b32_e32 v2, s1 +; GFX1250-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s4 +; GFX1250-NEXT: v_mov_b32_e32 v2, s5 ; GFX1250-NEXT: global_store_b32 v0, v1, s[2:3] scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_storecnt 0x0 ; GFX1250-NEXT: global_store_b32 v0, v2, s[2:3] scope:SCOPE_SYS @@ -404,10 +404,10 @@ define amdgpu_kernel void @byref_staggered_preload_arg(ptr addrspace(1) inreg %o ; ; GFX1250-LABEL: byref_staggered_preload_arg: ; GFX1250: ; %bb.0: -; GFX1250-NEXT: s_load_b64 s[0:1], s[0:1], 0x100 +; GFX1250-NEXT: s_load_b64 s[4:5], s[0:1], 0x100 ; GFX1250-NEXT: s_wait_kmcnt 0x0 -; GFX1250-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s0 -; GFX1250-NEXT: v_mov_b32_e32 v2, s1 +; GFX1250-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s4 +; GFX1250-NEXT: v_mov_b32_e32 v2, s5 ; GFX1250-NEXT: global_store_b32 v0, v1, s[2:3] scope:SCOPE_SYS ; GFX1250-NEXT: s_wait_storecnt 0x0 ; GFX1250-NEXT: global_store_b32 v0, v2, s[2:3] scope:SCOPE_SYS diff --git a/llvm/test/CodeGen/AMDGPU/regalloc-failure-overlapping-insert-assert.mir b/llvm/test/CodeGen/AMDGPU/regalloc-failure-overlapping-insert-assert.mir index b32e997..80afe7a 100644 --- a/llvm/test/CodeGen/AMDGPU/regalloc-failure-overlapping-insert-assert.mir +++ b/llvm/test/CodeGen/AMDGPU/regalloc-failure-overlapping-insert-assert.mir @@ -43,17 +43,17 @@ machineFunctionInfo: body: | bb.0: - INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 10 /* regdef */, implicit-def $agpr0 + INLINEASM &"; def $0", 1 /* sideeffect attdialect */, 2424842 /* regdef:AGPR_32 */, implicit-def $agpr0 %14:vgpr_32 = COPY killed $agpr0 - INLINEASM &"; def $0 $1 $2 $3 $4", 1 /* sideeffect attdialect */, 27394058 /* regdef:VReg_512 */, def %7, 13697034 /* regdef:VReg_256 */, def %8, 6225930 /* regdef:VReg_128 */, def %9, 4915210 /* regdef:VReg_96 */, def %10, 4915210 /* regdef:VReg_96 */, def %11 + INLINEASM &"; def $0 $1 $2 $3 $4", 1 /* sideeffect attdialect */, 40042506 /* regdef:VReg_512 */, def %7, 19464202 /* regdef:VReg_256 */, def %8, 7929866 /* regdef:VReg_128 */, def %9, 5963786 /* regdef:VReg_96 */, def %10, 5963786 /* regdef:VReg_96 */, def %11 INLINEASM &"; clobber", 1 /* sideeffect attdialect */, 12 /* clobber */, implicit-def dead early-clobber $agpr0_agpr1_agpr2_agpr3_agpr4_agpr5_agpr6_agpr7_agpr8_agpr9_agpr10_agpr11_agpr12_agpr13_agpr14_agpr15_agpr16_agpr17_agpr18_agpr19_agpr20_agpr21_agpr22_agpr23_agpr24_agpr25_agpr26_agpr27_agpr28_agpr29_agpr30_agpr31, 12 /* clobber */, implicit-def dead early-clobber $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 27394057 /* reguse:VReg_512 */, %7 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 13697033 /* reguse:VReg_256 */, %8 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 6225929 /* reguse:VReg_128 */, %9 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4915209 /* reguse:VReg_96 */, %10 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4915209 /* reguse:VReg_96 */, %11 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 40042505 /* reguse:VReg_512 */, %7 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 19464201 /* reguse:VReg_256 */, %8 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 7929865 /* reguse:VReg_128 */, %9 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 5963785 /* reguse:VReg_96 */, %10 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 5963785 /* reguse:VReg_96 */, %11 $agpr1 = COPY %14 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 9 /* reguse */, killed $agpr1 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 2424841 /* reguse:AGPR_32 */, killed $agpr1 SI_RETURN ... diff --git a/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-copy-from.mir b/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-copy-from.mir index 1b09f5d..ad490f8 100644 --- a/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-copy-from.mir +++ b/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-copy-from.mir @@ -41,9 +41,9 @@ body: | ; CHECK-NEXT: [[COPY1:%[0-9]+]]:av_64_align2 = COPY $vgpr0_vgpr1 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:av_64_align2 = COPY $vgpr2_vgpr3 ; CHECK-NEXT: [[GLOBAL_LOAD_DWORDX4_:%[0-9]+]]:areg_128_align2 = GLOBAL_LOAD_DWORDX4 [[COPY]], 0, 0, implicit $exec :: (load (s128), addrspace 1) - ; CHECK-NEXT: [[COPY3:%[0-9]+]]:areg_128_align2 = COPY [[GLOBAL_LOAD_DWORDX4_]] - ; CHECK-NEXT: [[V_MFMA_F64_4X4X4F64_e64_:%[0-9]+]]:areg_64_align2 = V_MFMA_F64_4X4X4F64_e64 [[COPY1]], [[COPY2]], [[COPY3]].sub0_sub1, 0, 0, 0, implicit $mode, implicit $exec - ; CHECK-NEXT: INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 3801097 /* reguse:AV_64_Align2 */, [[V_MFMA_F64_4X4X4F64_e64_]] + ; CHECK-NEXT: [[COPY3:%[0-9]+]]:vreg_128_align2 = COPY [[GLOBAL_LOAD_DWORDX4_]] + ; CHECK-NEXT: [[V_MFMA_F64_4X4X4F64_vgprcd_e64_:%[0-9]+]]:vreg_64_align2 = V_MFMA_F64_4X4X4F64_vgprcd_e64 [[COPY1]], [[COPY2]], [[COPY3]].sub0_sub1, 0, 0, 0, implicit $mode, implicit $exec + ; CHECK-NEXT: INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 3997705 /* reguse:VReg_64_Align2 */, [[V_MFMA_F64_4X4X4F64_vgprcd_e64_]] ; CHECK-NEXT: SI_RETURN %0:vreg_64_align2 = COPY $vgpr4_vgpr5 %1:av_64_align2 = COPY $vgpr0_vgpr1 @@ -51,7 +51,7 @@ body: | %3:areg_128_align2 = GLOBAL_LOAD_DWORDX4 %0, 0, 0, implicit $exec :: (load (s128), addrspace 1) %4:vreg_128_align2 = COPY %3 %5:vreg_64_align2 = V_MFMA_F64_4X4X4F64_vgprcd_e64 %1, %2, %4.sub0_sub1, 0, 0, 0, implicit $mode, implicit $exec - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 3801097 /* reguse:VReg_64_Align2 */, %5 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 3997705 /* reguse:VReg_64_Align2 */, %5 SI_RETURN ... diff --git a/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-subreg-src2-chain.mir b/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-subreg-src2-chain.mir index 4c2ea2f..dcf3b8b 100644 --- a/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-subreg-src2-chain.mir +++ b/llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-subreg-src2-chain.mir @@ -79,7 +79,7 @@ body: | ; CHECK-NEXT: dead %other_use:vreg_64_align2 = COPY [[V_MFMA_F64_4X4X4F64_e64_]].sub0_sub1 ; CHECK-NEXT: [[V_MFMA_F64_4X4X4F64_e64_1:%[0-9]+]]:areg_64_align2 = V_MFMA_F64_4X4X4F64_e64 [[COPY1]], [[COPY2]], [[V_MFMA_F64_4X4X4F64_e64_]].sub0_sub1, 0, 0, 0, implicit $mode, implicit $exec ; CHECK-NEXT: [[COPY3:%[0-9]+]]:areg_64_align2 = COPY [[V_MFMA_F64_4X4X4F64_e64_1]] - ; CHECK-NEXT: INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4521993 /* reguse:VS_64_with_sub0_in_VS_32_Lo128 */, [[COPY3]] + ; CHECK-NEXT: INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4915209 /* reguse:AReg_64_Align2 */, [[COPY3]] ; CHECK-NEXT: GLOBAL_STORE_DWORDX2 [[COPY]], [[COPY3]], 0, 0, implicit $exec :: (store (s64), addrspace 1) ; CHECK-NEXT: SI_RETURN %0:vreg_64_align2 = COPY $vgpr4_vgpr5 @@ -90,7 +90,7 @@ body: | %other_use:vreg_64_align2 = COPY %4.sub0_sub1 %5:vreg_64_align2 = V_MFMA_F64_4X4X4F64_vgprcd_e64 %1, %2, %4.sub0_sub1, 0, 0, 0, implicit $mode, implicit $exec %6:areg_64_align2 = COPY %5 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4521993 /* reguse:AReg_64_Align2 */, %6:areg_64_align2 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4915209 /* reguse:AReg_64_Align2 */, %6:areg_64_align2 GLOBAL_STORE_DWORDX2 %0, %6, 0, 0, implicit $exec :: (store (s64), addrspace 1) SI_RETURN ... @@ -114,7 +114,7 @@ body: | ; CHECK-NEXT: undef [[V_MFMA_F64_4X4X4F64_e64_1:%[0-9]+]].sub0_sub1:areg_128_align2 = V_MFMA_F64_4X4X4F64_e64 [[COPY1]], [[COPY2]], [[V_MFMA_F64_4X4X4F64_e64_]], 0, 0, 0, implicit $mode, implicit $exec ; CHECK-NEXT: [[V_MFMA_F64_4X4X4F64_e64_2:%[0-9]+]]:areg_64_align2 = V_MFMA_F64_4X4X4F64_e64 [[COPY1]], [[COPY2]], [[V_MFMA_F64_4X4X4F64_e64_1]].sub0_sub1, 0, 0, 0, implicit $mode, implicit $exec ; CHECK-NEXT: [[COPY3:%[0-9]+]]:areg_64_align2 = COPY [[V_MFMA_F64_4X4X4F64_e64_2]] - ; CHECK-NEXT: INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4521993 /* reguse:VS_64_with_sub0_in_VS_32_Lo128 */, [[COPY3]] + ; CHECK-NEXT: INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4915209 /* reguse:AReg_64_Align2 */, [[COPY3]] ; CHECK-NEXT: GLOBAL_STORE_DWORDX2 [[COPY]], [[COPY3]], 0, 0, implicit $exec :: (store (s64), addrspace 1) ; CHECK-NEXT: SI_RETURN %0:vreg_64_align2 = COPY $vgpr4_vgpr5 @@ -126,7 +126,7 @@ body: | undef %5.sub0_sub1:vreg_128_align2 = V_MFMA_F64_4X4X4F64_vgprcd_e64 %1, %2, %4, 0, 0, 0, implicit $mode, implicit $exec %6:vreg_64_align2 = V_MFMA_F64_4X4X4F64_vgprcd_e64 %1, %2, %5.sub0_sub1, 0, 0, 0, implicit $mode, implicit $exec %7:areg_64_align2 = COPY %6 - INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4521993 /* reguse:AReg_64_Align2 */, %7 + INLINEASM &"; use $0", 1 /* sideeffect attdialect */, 4915209 /* reguse:AReg_64_Align2 */, %7 GLOBAL_STORE_DWORDX2 %0, %7, 0, 0, implicit $exec :: (store (s64), addrspace 1) SI_RETURN diff --git a/llvm/test/CodeGen/AMDGPU/scale-offset-flat.ll b/llvm/test/CodeGen/AMDGPU/scale-offset-flat.ll index 335d58c..a18847b 100644 --- a/llvm/test/CodeGen/AMDGPU/scale-offset-flat.ll +++ b/llvm/test/CodeGen/AMDGPU/scale-offset-flat.ll @@ -324,11 +324,9 @@ define amdgpu_ps <2 x float> @flat_atomicrmw_b64_rtn_idxprom(ptr align 8 inreg % ; SDAG-LABEL: flat_atomicrmw_b64_rtn_idxprom: ; SDAG: ; %bb.0: ; %entry ; SDAG-NEXT: v_ashrrev_i32_e32 v1, 31, v0 -; SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) +; SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; SDAG-NEXT: v_lshl_add_u64 v[2:3], v[0:1], 3, s[0:1] -; SDAG-NEXT: s_mov_b32 s0, src_flat_scratch_base_hi -; SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) -; SDAG-NEXT: v_xor_b32_e32 v0, s0, v3 +; SDAG-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v3 ; SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) ; SDAG-NEXT: v_cmp_lt_u32_e32 vcc_lo, 0x3ffffff, v0 ; SDAG-NEXT: ; implicit-def: $vgpr0_vgpr1 @@ -350,10 +348,9 @@ define amdgpu_ps <2 x float> @flat_atomicrmw_b64_rtn_idxprom(ptr align 8 inreg % ; SDAG-NEXT: s_and_not1_saveexec_b32 s0, s0 ; SDAG-NEXT: s_cbranch_execz .LBB21_2 ; SDAG-NEXT: .LBB21_4: ; %atomicrmw.private -; SDAG-NEXT: s_mov_b32 s1, src_flat_scratch_base_lo ; SDAG-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3] ; SDAG-NEXT: s_wait_loadcnt_dscnt 0x0 -; SDAG-NEXT: v_subrev_nc_u32_e32 v0, s1, v2 +; SDAG-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v2 ; SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1) ; SDAG-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; SDAG-NEXT: scratch_load_b64 v[0:1], v4, off @@ -367,12 +364,12 @@ define amdgpu_ps <2 x float> @flat_atomicrmw_b64_rtn_idxprom(ptr align 8 inreg % ; ; GISEL-LABEL: flat_atomicrmw_b64_rtn_idxprom: ; GISEL: ; %bb.0: ; %entry -; GISEL-NEXT: v_dual_mov_b32 v2, v0 :: v_dual_mov_b32 v0, src_flat_scratch_base_hi +; GISEL-NEXT: v_mov_b32_e32 v2, v0 ; GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GISEL-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GISEL-NEXT: v_lshl_add_u64 v[4:5], v[2:3], 3, s[0:1] ; GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GISEL-NEXT: v_xor_b32_e32 v0, v5, v0 +; GISEL-NEXT: v_xor_b32_e32 v0, src_flat_scratch_base_hi, v5 ; GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 0x4000000, v0 ; GISEL-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GISEL-NEXT: s_and_saveexec_b32 s2, vcc_lo @@ -394,11 +391,10 @@ define amdgpu_ps <2 x float> @flat_atomicrmw_b64_rtn_idxprom(ptr align 8 inreg % ; GISEL-NEXT: s_and_not1_saveexec_b32 s0, s2 ; GISEL-NEXT: s_cbranch_execz .LBB21_2 ; GISEL-NEXT: .LBB21_4: ; %atomicrmw.private -; GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 -; GISEL-NEXT: v_mov_b32_e32 v0, src_flat_scratch_base_lo ; GISEL-NEXT: v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5] -; GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) -; GISEL-NEXT: v_sub_nc_u32_e32 v0, v4, v0 +; GISEL-NEXT: s_wait_loadcnt_dscnt 0x0 +; GISEL-NEXT: v_subrev_nc_u32_e32 v0, src_flat_scratch_base_lo, v4 +; GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GISEL-NEXT: v_cndmask_b32_e32 v4, -1, v0, vcc_lo ; GISEL-NEXT: scratch_load_b64 v[0:1], v4, off ; GISEL-NEXT: s_wait_loadcnt 0x0 diff --git a/llvm/test/CodeGen/AMDGPU/scale-offset-smem.ll b/llvm/test/CodeGen/AMDGPU/scale-offset-smem.ll index b5bb68e..e0ea08d 100644 --- a/llvm/test/CodeGen/AMDGPU/scale-offset-smem.ll +++ b/llvm/test/CodeGen/AMDGPU/scale-offset-smem.ll @@ -97,9 +97,9 @@ entry: define amdgpu_ps <2 x float> @s_load_b64_idxprom(ptr addrspace(4) align 4 inreg %p, i32 inreg %idx) { ; GCN-LABEL: s_load_b64_idxprom: ; GCN: ; %bb.0: ; %entry -; GCN-NEXT: s_load_b64 s[0:1], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b64 s[4:5], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 ; GCN-NEXT: ; return to shader part epilog entry: %idxprom = zext i32 %idx to i64 @@ -111,10 +111,10 @@ entry: define amdgpu_ps <3 x float> @s_load_b96_idxprom(ptr addrspace(4) align 4 inreg %p, i32 inreg %idx) { ; GCN-LABEL: s_load_b96_idxprom: ; GCN: ; %bb.0: ; %entry -; GCN-NEXT: s_load_b96 s[0:2], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b96 s[4:6], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_mov_b32_e32 v2, s2 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_mov_b32_e32 v2, s6 ; GCN-NEXT: ; return to shader part epilog entry: %idxprom = zext i32 %idx to i64 @@ -126,10 +126,10 @@ entry: define amdgpu_ps <4 x float> @s_load_b128_idxprom(ptr addrspace(4) align 4 inreg %p, i32 inreg %idx) { ; GCN-LABEL: s_load_b128_idxprom: ; GCN: ; %bb.0: ; %entry -; GCN-NEXT: s_load_b128 s[0:3], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b128 s[4:7], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_dual_mov_b32 v2, s6 :: v_dual_mov_b32 v3, s7 ; GCN-NEXT: ; return to shader part epilog entry: %idxprom = zext i32 %idx to i64 @@ -141,12 +141,12 @@ entry: define amdgpu_ps <8 x float> @s_load_b256_idxprom(ptr addrspace(4) align 4 inreg %p, i32 inreg %idx) { ; GCN-LABEL: s_load_b256_idxprom: ; GCN: ; %bb.0: ; %entry -; GCN-NEXT: s_load_b256 s[0:7], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b256 s[4:11], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3 -; GCN-NEXT: v_dual_mov_b32 v4, s4 :: v_dual_mov_b32 v5, s5 -; GCN-NEXT: v_dual_mov_b32 v6, s6 :: v_dual_mov_b32 v7, s7 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_dual_mov_b32 v2, s6 :: v_dual_mov_b32 v3, s7 +; GCN-NEXT: v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, s9 +; GCN-NEXT: v_dual_mov_b32 v6, s10 :: v_dual_mov_b32 v7, s11 ; GCN-NEXT: ; return to shader part epilog entry: %idxprom = zext i32 %idx to i64 @@ -158,16 +158,16 @@ entry: define amdgpu_ps <16 x float> @s_load_b512_idxprom(ptr addrspace(4) align 4 inreg %p, i32 inreg %idx) { ; GCN-LABEL: s_load_b512_idxprom: ; GCN: ; %bb.0: ; %entry -; GCN-NEXT: s_load_b512 s[0:15], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b512 s[4:19], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3 -; GCN-NEXT: v_dual_mov_b32 v4, s4 :: v_dual_mov_b32 v5, s5 -; GCN-NEXT: v_dual_mov_b32 v6, s6 :: v_dual_mov_b32 v7, s7 -; GCN-NEXT: v_dual_mov_b32 v8, s8 :: v_dual_mov_b32 v9, s9 -; GCN-NEXT: v_dual_mov_b32 v10, s10 :: v_dual_mov_b32 v11, s11 -; GCN-NEXT: v_dual_mov_b32 v12, s12 :: v_dual_mov_b32 v13, s13 -; GCN-NEXT: v_dual_mov_b32 v14, s14 :: v_dual_mov_b32 v15, s15 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_dual_mov_b32 v2, s6 :: v_dual_mov_b32 v3, s7 +; GCN-NEXT: v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, s9 +; GCN-NEXT: v_dual_mov_b32 v6, s10 :: v_dual_mov_b32 v7, s11 +; GCN-NEXT: v_dual_mov_b32 v8, s12 :: v_dual_mov_b32 v9, s13 +; GCN-NEXT: v_dual_mov_b32 v10, s14 :: v_dual_mov_b32 v11, s15 +; GCN-NEXT: v_dual_mov_b32 v12, s16 :: v_dual_mov_b32 v13, s17 +; GCN-NEXT: v_dual_mov_b32 v14, s18 :: v_dual_mov_b32 v15, s19 ; GCN-NEXT: ; return to shader part epilog entry: %idxprom = zext i32 %idx to i64 @@ -275,11 +275,11 @@ entry: define amdgpu_ps <2 x float> @s_load_b64_idxprom_range(ptr addrspace(4) align 4 inreg %p) { ; GCN-LABEL: s_load_b64_idxprom_range: ; GCN: ; %bb.0: ; %entry -; GCN-NEXT: s_load_b32 s2, s[0:1], 0x0 +; GCN-NEXT: s_load_b32 s4, s[0:1], 0x0 ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: s_load_b64 s[0:1], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b64 s[2:3], s[0:1], s4 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 +; GCN-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 ; GCN-NEXT: ; return to shader part epilog entry: %idx = load i32, ptr addrspace(4) %p, align 4, !range !0 @@ -294,10 +294,10 @@ define amdgpu_ps <3 x float> @s_load_b96_idxprom_range(ptr addrspace(4) align 4 ; GCN: ; %bb.0: ; %entry ; GCN-NEXT: s_load_b32 s2, s[0:1], 0x0 ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: s_load_b96 s[0:2], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b96 s[4:6], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_mov_b32_e32 v2, s2 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_mov_b32_e32 v2, s6 ; GCN-NEXT: ; return to shader part epilog entry: %idx = load i32, ptr addrspace(4) %p, align 4, !range !0 @@ -312,10 +312,10 @@ define amdgpu_ps <4 x float> @s_load_b128_idxprom_range(ptr addrspace(4) align 4 ; GCN: ; %bb.0: ; %entry ; GCN-NEXT: s_load_b32 s2, s[0:1], 0x0 ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: s_load_b128 s[0:3], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b128 s[4:7], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_dual_mov_b32 v2, s6 :: v_dual_mov_b32 v3, s7 ; GCN-NEXT: ; return to shader part epilog entry: %idx = load i32, ptr addrspace(4) %p, align 4, !range !0 @@ -330,12 +330,12 @@ define amdgpu_ps <8 x float> @s_load_b256_idxprom_range(ptr addrspace(4) align 4 ; GCN: ; %bb.0: ; %entry ; GCN-NEXT: s_load_b32 s2, s[0:1], 0x0 ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: s_load_b256 s[0:7], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b256 s[4:11], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3 -; GCN-NEXT: v_dual_mov_b32 v4, s4 :: v_dual_mov_b32 v5, s5 -; GCN-NEXT: v_dual_mov_b32 v6, s6 :: v_dual_mov_b32 v7, s7 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_dual_mov_b32 v2, s6 :: v_dual_mov_b32 v3, s7 +; GCN-NEXT: v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, s9 +; GCN-NEXT: v_dual_mov_b32 v6, s10 :: v_dual_mov_b32 v7, s11 ; GCN-NEXT: ; return to shader part epilog entry: %idx = load i32, ptr addrspace(4) %p, align 4, !range !0 @@ -350,16 +350,16 @@ define amdgpu_ps <16 x float> @s_load_b512_idxprom_range(ptr addrspace(4) align ; GCN: ; %bb.0: ; %entry ; GCN-NEXT: s_load_b32 s2, s[0:1], 0x0 ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: s_load_b512 s[0:15], s[0:1], s2 offset:0x0 scale_offset +; GCN-NEXT: s_load_b512 s[4:19], s[0:1], s2 offset:0x0 scale_offset ; GCN-NEXT: s_wait_kmcnt 0x0 -; GCN-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 -; GCN-NEXT: v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3 -; GCN-NEXT: v_dual_mov_b32 v4, s4 :: v_dual_mov_b32 v5, s5 -; GCN-NEXT: v_dual_mov_b32 v6, s6 :: v_dual_mov_b32 v7, s7 -; GCN-NEXT: v_dual_mov_b32 v8, s8 :: v_dual_mov_b32 v9, s9 -; GCN-NEXT: v_dual_mov_b32 v10, s10 :: v_dual_mov_b32 v11, s11 -; GCN-NEXT: v_dual_mov_b32 v12, s12 :: v_dual_mov_b32 v13, s13 -; GCN-NEXT: v_dual_mov_b32 v14, s14 :: v_dual_mov_b32 v15, s15 +; GCN-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GCN-NEXT: v_dual_mov_b32 v2, s6 :: v_dual_mov_b32 v3, s7 +; GCN-NEXT: v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, s9 +; GCN-NEXT: v_dual_mov_b32 v6, s10 :: v_dual_mov_b32 v7, s11 +; GCN-NEXT: v_dual_mov_b32 v8, s12 :: v_dual_mov_b32 v9, s13 +; GCN-NEXT: v_dual_mov_b32 v10, s14 :: v_dual_mov_b32 v11, s15 +; GCN-NEXT: v_dual_mov_b32 v12, s16 :: v_dual_mov_b32 v13, s17 +; GCN-NEXT: v_dual_mov_b32 v14, s18 :: v_dual_mov_b32 v15, s19 ; GCN-NEXT: ; return to shader part epilog entry: %idx = load i32, ptr addrspace(4) %p, align 4, !range !0 diff --git a/llvm/test/CodeGen/AMDGPU/v_ashr_pk.ll b/llvm/test/CodeGen/AMDGPU/v_ashr_pk.ll index f2ecfe8..3d74b17 100644 --- a/llvm/test/CodeGen/AMDGPU/v_ashr_pk.ll +++ b/llvm/test/CodeGen/AMDGPU/v_ashr_pk.ll @@ -17,16 +17,16 @@ define amdgpu_kernel void @v_ashr_pk_i8_i32(ptr addrspace(1) %out, i32 %src0, i3 ; ; GFX1250-LABEL: v_ashr_pk_i8_i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b96 s[0:2], s[4:5], 0x2c -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x24 ; GFX1250-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_and_b32 s2, s2, 31 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_mov_b32_e32 v0, s2 ; GFX1250-NEXT: v_ashr_pk_i8_i32 v0, s0, s1, v0 -; GFX1250-NEXT: global_store_b16 v1, v0, s[4:5] +; GFX1250-NEXT: global_store_b16 v1, v0, s[6:7] ; GFX1250-NEXT: s_endpgm %insert.0 = insertelement <2 x i32> poison, i32 %src0, i64 0 %build_vector = insertelement <2 x i32> %insert.0, i32 %src1, i64 1 @@ -58,16 +58,16 @@ define amdgpu_kernel void @v_ashr_pk_u8_i32(ptr addrspace(1) %out, i32 %src0, i3 ; ; GFX1250-LABEL: v_ashr_pk_u8_i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_clause 0x1 ; GFX1250-NEXT: s_load_b96 s[0:2], s[4:5], 0x2c -; GFX1250-NEXT: s_wait_xcnt 0x0 -; GFX1250-NEXT: s_load_b64 s[4:5], s[4:5], 0x24 +; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x24 ; GFX1250-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: s_and_b32 s2, s2, 31 ; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_mov_b32_e32 v0, s2 ; GFX1250-NEXT: v_ashr_pk_u8_i32 v0, s0, s1, v0 -; GFX1250-NEXT: global_store_b16 v1, v0, s[4:5] +; GFX1250-NEXT: global_store_b16 v1, v0, s[6:7] ; GFX1250-NEXT: s_endpgm %insert.0 = insertelement <2 x i32> poison, i32 %src0, i64 0 %build_vector = insertelement <2 x i32> %insert.0, i32 %src1, i64 1 diff --git a/llvm/test/CodeGen/AMDGPU/workgroup-id-in-arch-sgprs.ll b/llvm/test/CodeGen/AMDGPU/workgroup-id-in-arch-sgprs.ll index a392692..6636eb5 100644 --- a/llvm/test/CodeGen/AMDGPU/workgroup-id-in-arch-sgprs.ll +++ b/llvm/test/CodeGen/AMDGPU/workgroup-id-in-arch-sgprs.ll @@ -211,38 +211,39 @@ define amdgpu_kernel void @workgroup_id_xyz(ptr addrspace(1) %ptrx, ptr addrspac ; GFX1250-SDAG-LABEL: workgroup_id_xyz: ; GFX1250-SDAG: ; %bb.0: ; GFX1250-SDAG-NEXT: s_bfe_u32 s0, ttmp6, 0x40014 -; GFX1250-SDAG-NEXT: s_lshr_b32 s6, ttmp7, 16 -; GFX1250-SDAG-NEXT: s_add_co_i32 s7, s0, 1 +; GFX1250-SDAG-NEXT: s_lshr_b32 s8, ttmp7, 16 +; GFX1250-SDAG-NEXT: s_add_co_i32 s9, s0, 1 +; GFX1250-SDAG-NEXT: s_clause 0x1 ; GFX1250-SDAG-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 +; GFX1250-SDAG-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 ; GFX1250-SDAG-NEXT: s_wait_xcnt 0x0 -; GFX1250-SDAG-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 +; GFX1250-SDAG-NEXT: s_mul_i32 s4, s8, s9 ; GFX1250-SDAG-NEXT: s_bfe_u32 s9, ttmp6, 0x40010 -; GFX1250-SDAG-NEXT: s_mul_i32 s7, s6, s7 -; GFX1250-SDAG-NEXT: s_bfe_u32 s8, ttmp6, 0x40008 +; GFX1250-SDAG-NEXT: s_bfe_u32 s5, ttmp6, 0x40008 ; GFX1250-SDAG-NEXT: s_and_b32 s10, ttmp7, 0xffff ; GFX1250-SDAG-NEXT: s_add_co_i32 s9, s9, 1 ; GFX1250-SDAG-NEXT: s_bfe_u32 s11, ttmp6, 0x4000c -; GFX1250-SDAG-NEXT: s_add_co_i32 s8, s8, s7 -; GFX1250-SDAG-NEXT: s_mul_i32 s7, s10, s9 +; GFX1250-SDAG-NEXT: s_add_co_i32 s5, s5, s4 +; GFX1250-SDAG-NEXT: s_mul_i32 s4, s10, s9 ; GFX1250-SDAG-NEXT: s_bfe_u32 s9, ttmp6, 0x40004 ; GFX1250-SDAG-NEXT: s_add_co_i32 s11, s11, 1 -; GFX1250-SDAG-NEXT: s_add_co_i32 s9, s9, s7 -; GFX1250-SDAG-NEXT: s_and_b32 s7, ttmp6, 15 +; GFX1250-SDAG-NEXT: s_add_co_i32 s9, s9, s4 +; GFX1250-SDAG-NEXT: s_and_b32 s4, ttmp6, 15 ; GFX1250-SDAG-NEXT: s_mul_i32 s11, ttmp9, s11 ; GFX1250-SDAG-NEXT: s_getreg_b32 s12, hwreg(HW_REG_IB_STS2, 6, 4) -; GFX1250-SDAG-NEXT: s_add_co_i32 s7, s7, s11 +; GFX1250-SDAG-NEXT: s_add_co_i32 s4, s4, s11 ; GFX1250-SDAG-NEXT: s_cmp_eq_u32 s12, 0 -; GFX1250-SDAG-NEXT: s_cselect_b32 s7, ttmp9, s7 +; GFX1250-SDAG-NEXT: s_cselect_b32 s4, ttmp9, s4 ; GFX1250-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s7 -; GFX1250-SDAG-NEXT: s_cselect_b32 s7, s10, s9 -; GFX1250-SDAG-NEXT: s_cselect_b32 s6, s6, s8 -; GFX1250-SDAG-NEXT: v_dual_mov_b32 v2, s7 :: v_dual_mov_b32 v3, s6 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s4 +; GFX1250-SDAG-NEXT: s_cselect_b32 s4, s10, s9 +; GFX1250-SDAG-NEXT: s_cselect_b32 s5, s8, s5 +; GFX1250-SDAG-NEXT: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s5 ; GFX1250-SDAG-NEXT: s_wait_kmcnt 0x0 ; GFX1250-SDAG-NEXT: s_clause 0x2 ; GFX1250-SDAG-NEXT: global_store_b32 v0, v1, s[0:1] ; GFX1250-SDAG-NEXT: global_store_b32 v0, v2, s[2:3] -; GFX1250-SDAG-NEXT: global_store_b32 v0, v3, s[4:5] +; GFX1250-SDAG-NEXT: global_store_b32 v0, v3, s[6:7] ; GFX1250-SDAG-NEXT: s_endpgm ; ; GFX1250-GISEL-LABEL: workgroup_id_xyz: @@ -250,39 +251,40 @@ define amdgpu_kernel void @workgroup_id_xyz(ptr addrspace(1) %ptrx, ptr addrspac ; GFX1250-GISEL-NEXT: s_bfe_u32 s0, ttmp6, 0x4000c ; GFX1250-GISEL-NEXT: s_and_b32 s1, ttmp6, 15 ; GFX1250-GISEL-NEXT: s_add_co_i32 s0, s0, 1 -; GFX1250-GISEL-NEXT: s_getreg_b32 s6, hwreg(HW_REG_IB_STS2, 6, 4) +; GFX1250-GISEL-NEXT: s_getreg_b32 s8, hwreg(HW_REG_IB_STS2, 6, 4) ; GFX1250-GISEL-NEXT: s_mul_i32 s0, ttmp9, s0 ; GFX1250-GISEL-NEXT: v_mov_b32_e32 v1, 0 ; GFX1250-GISEL-NEXT: s_add_co_i32 s1, s1, s0 -; GFX1250-GISEL-NEXT: s_cmp_eq_u32 s6, 0 -; GFX1250-GISEL-NEXT: s_cselect_b32 s7, ttmp9, s1 +; GFX1250-GISEL-NEXT: s_cmp_eq_u32 s8, 0 +; GFX1250-GISEL-NEXT: s_cselect_b32 s9, ttmp9, s1 ; GFX1250-GISEL-NEXT: s_bfe_u32 s0, ttmp6, 0x40010 -; GFX1250-GISEL-NEXT: s_and_b32 s8, ttmp7, 0xffff +; GFX1250-GISEL-NEXT: s_and_b32 s10, ttmp7, 0xffff ; GFX1250-GISEL-NEXT: s_add_co_i32 s0, s0, 1 -; GFX1250-GISEL-NEXT: s_bfe_u32 s9, ttmp6, 0x40004 -; GFX1250-GISEL-NEXT: s_mul_i32 s10, s8, s0 +; GFX1250-GISEL-NEXT: s_bfe_u32 s11, ttmp6, 0x40004 +; GFX1250-GISEL-NEXT: s_mul_i32 s12, s10, s0 +; GFX1250-GISEL-NEXT: s_clause 0x1 ; GFX1250-GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x0 +; GFX1250-GISEL-NEXT: s_load_b64 s[6:7], s[4:5], 0x10 +; GFX1250-GISEL-NEXT: s_add_co_i32 s11, s11, s12 +; GFX1250-GISEL-NEXT: s_cmp_eq_u32 s8, 0 +; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, s9 ; GFX1250-GISEL-NEXT: s_wait_xcnt 0x0 -; GFX1250-GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x10 -; GFX1250-GISEL-NEXT: s_add_co_i32 s9, s9, s10 -; GFX1250-GISEL-NEXT: s_cmp_eq_u32 s6, 0 -; GFX1250-GISEL-NEXT: v_mov_b32_e32 v0, s7 -; GFX1250-GISEL-NEXT: s_cselect_b32 s8, s8, s9 -; GFX1250-GISEL-NEXT: s_bfe_u32 s9, ttmp6, 0x40014 +; GFX1250-GISEL-NEXT: s_cselect_b32 s4, s10, s11 +; GFX1250-GISEL-NEXT: s_bfe_u32 s5, ttmp6, 0x40014 ; GFX1250-GISEL-NEXT: s_lshr_b32 s10, ttmp7, 16 -; GFX1250-GISEL-NEXT: s_add_co_i32 s9, s9, 1 +; GFX1250-GISEL-NEXT: s_add_co_i32 s5, s5, 1 ; GFX1250-GISEL-NEXT: s_bfe_u32 s11, ttmp6, 0x40008 -; GFX1250-GISEL-NEXT: s_mul_i32 s9, s10, s9 +; GFX1250-GISEL-NEXT: s_mul_i32 s5, s10, s5 ; GFX1250-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) -; GFX1250-GISEL-NEXT: s_add_co_i32 s11, s11, s9 -; GFX1250-GISEL-NEXT: s_cmp_eq_u32 s6, 0 -; GFX1250-GISEL-NEXT: s_cselect_b32 s6, s10, s11 -; GFX1250-GISEL-NEXT: v_dual_mov_b32 v2, s8 :: v_dual_mov_b32 v3, s6 +; GFX1250-GISEL-NEXT: s_add_co_i32 s11, s11, s5 +; GFX1250-GISEL-NEXT: s_cmp_eq_u32 s8, 0 +; GFX1250-GISEL-NEXT: s_cselect_b32 s5, s10, s11 +; GFX1250-GISEL-NEXT: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s5 ; GFX1250-GISEL-NEXT: s_wait_kmcnt 0x0 ; GFX1250-GISEL-NEXT: s_clause 0x2 ; GFX1250-GISEL-NEXT: global_store_b32 v1, v0, s[0:1] ; GFX1250-GISEL-NEXT: global_store_b32 v1, v2, s[2:3] -; GFX1250-GISEL-NEXT: global_store_b32 v1, v3, s[4:5] +; GFX1250-GISEL-NEXT: global_store_b32 v1, v3, s[6:7] ; GFX1250-GISEL-NEXT: s_endpgm ; GFX12-LABEL: workgroup_id_xyz: ; GFX12: ; %bb.0: diff --git a/llvm/test/CodeGen/Hexagon/inst_setcc_uno_uo.ll b/llvm/test/CodeGen/Hexagon/inst_setcc_uno_uo.ll new file mode 100644 index 0000000..8b121c5 --- /dev/null +++ b/llvm/test/CodeGen/Hexagon/inst_setcc_uno_uo.ll @@ -0,0 +1,93 @@ +;; RUN: llc --mtriple=hexagon -mattr=+hvxv79,+hvx-length128b %s -o - | FileCheck %s + +define dso_local void @store_isnan_f32(ptr %a, ptr %b, ptr %isnan_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw float, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw float, ptr %b, i32 0 + %0 = load <32 x float>, ptr %arrayidx_a, align 4 + %1 = load <32 x float>, ptr %arrayidx_b, align 4 + %.vectorized = fcmp uno <32 x float> %0, %1 + %.LS.instance = zext <32 x i1> %.vectorized to <32 x i32> + %arrayidx1 = getelementptr inbounds nuw i32, ptr %isnan_cmp, i32 0 + store <32 x i32> %.LS.instance, ptr %arrayidx1, align 4 + ret void +} + +; CHECK: store_isnan_f32 +; CHECK: [[RONE32:r[0-9]+]] = #1 +; CHECK: [[VOP2_F32:v[0-9]+]] = vxor([[VOP2_F32]],[[VOP2_F32]]) +; CHECK: [[VOP1_F32:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES32:v[0-9]+]] = vsplat([[RONE32]]) +; CHECK: [[Q1_F32:q[0-9]+]] = vcmp.eq([[VOP1_F32]].w,[[VOP1_F32]].w) +; CHECK: [[VOP3_F32:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_F32]] &= vcmp.eq([[VOP3_F32]].w,[[VOP3_F32]].w) +; CHECK: [[VOUT_F32:v[0-9]+]] = vmux([[Q1_F32]],[[VOP2_F32]],[[VONES32]]) +; CHECK: vmemu(r2+#0) = [[VOUT_F32]] + +define dso_local void @store_isnan_f16(ptr %a, ptr %b, ptr %isnan_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw half, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw half, ptr %b, i32 0 + %0 = load <64 x half>, ptr %arrayidx_a, align 2 + %1 = load <64 x half>, ptr %arrayidx_b, align 2 + %.vectorized = fcmp uno <64 x half> %0, %1 + %conv.LS.instance = zext <64 x i1> %.vectorized to <64 x i16> + %arrayidx1 = getelementptr inbounds nuw i16, ptr %isnan_cmp, i32 0 + store <64 x i16> %conv.LS.instance, ptr %arrayidx1, align 2 + ret void +} +; CHECK-LABEL: store_isnan_f16 +; CHECK: [[RONE16:r[0-9]+]] = #1 +; CHECK: [[VOP2_F16:v[0-9]+]] = vxor([[VOP2_F16]],[[VOP2_F16]]) +; CHECK: [[VOP1_F16:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES16:v[0-9]+]].h = vsplat([[RONE16]]) +; CHECK: [[Q1_F16:q[0-9]+]] = vcmp.eq([[VOP1_F16]].h,[[VOP1_F16]].h) +; CHECK: [[VOP3_F16:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_F16]] &= vcmp.eq([[VOP3_F16]].h,[[VOP3_F16]].h) +; CHECK: [[VOUT_F16:v[0-9]+]] = vmux([[Q1_F16]],[[VOP2_F16]],[[VONES16]]) +; CHECK: vmemu(r2+#0) = [[VOUT_F32]] + +define dso_local void @store_isordered_f32(ptr %a, ptr %b, ptr %isordered_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw float, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw float, ptr %b, i32 0 + %0 = load <32 x float>, ptr %arrayidx_a, align 4 + %1 = load <32 x float>, ptr %arrayidx_b, align 4 + %.vectorized = fcmp ord <32 x float> %0, %1 + %.LS.instance = zext <32 x i1> %.vectorized to <32 x i32> + %arrayidx1 = getelementptr inbounds nuw i32, ptr %isordered_cmp, i32 0 + store <32 x i32> %.LS.instance, ptr %arrayidx1, align 4 + ret void +} +; CHECK-LABEL: store_isordered_f32 +; CHECK: [[VOP2_ORD_F32:v[0-9]+]] = vxor([[VOP2_ORD_F32]],[[VOP2_ORD_F32]]) +; CHECK: [[VOP1_ORD_F32:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES_ORD_F32:v[0-9]+]] = vsplat([[RONE32]]) +; CHECK: [[Q1_ORD_F32:q[0-9]+]] = vcmp.eq([[VOP1_ORD_F32]].w,[[VOP1_ORD_F32]].w) +; CHECK: [[VOP3_ORD_F32:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_ORD_F32]] &= vcmp.eq([[VOP3_ORD_F32]].w,[[VOP3_ORD_F32]].w) +; CHECK: [[VOUT_ORD_F32:v[0-9]+]] = vmux([[Q1_ORD_F32]],[[VONES_ORD_F32]],[[VOP2_ORD_F32]]) +; CHECK: vmemu(r2+#0) = [[VOUT_ORD_F32]] + + +define dso_local void @store_isordered_f16(ptr %a, ptr %b, ptr %isordered_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw half, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw half, ptr %b, i32 0 + %0 = load <64 x half>, ptr %arrayidx_a, align 2 + %1 = load <64 x half>, ptr %arrayidx_b, align 2 + %.vectorized = fcmp ord <64 x half> %0, %1 + %conv.LS.instance = zext <64 x i1> %.vectorized to <64 x i16> + %arrayidx1 = getelementptr inbounds nuw i16, ptr %isordered_cmp, i32 0 + store <64 x i16> %conv.LS.instance, ptr %arrayidx1, align 2 + ret void +} +; CHECK-LABEL: store_isordered_f16 +; CHECK: [[VOP2_ORD_F16:v[0-9]+]] = vxor([[VOP2_ORD_F16]],[[VOP2_ORD_F16]]) +; CHECK: [[VOP1_ORD_F16:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES_ORD_F16:v[0-9]+]].h = vsplat([[RONE16]]) +; CHECK: [[Q1_ORD_F16:q[0-9]+]] = vcmp.eq([[VOP1_ORD_F16]].h,[[VOP1_ORD_F16]].h) +; CHECK: [[VOP3_ORD_F16:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_ORD_F16]] &= vcmp.eq([[VOP3_ORD_F16]].h,[[VOP3_ORD_F16]].h) +; CHECK: [[VOUT_ORD_F16:v[0-9]+]] = vmux([[Q1_ORD_F16]],[[VONES_ORD_F16]],[[VOP2_ORD_F16]]) +; CHECK: vmemu(r2+#0) = [[VOUT_ORD_F16]] diff --git a/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll b/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll index 5fa5023..fe0f7dd 100644 --- a/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll +++ b/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll @@ -1,3 +1,4 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6 ; RUN: llc -mtriple=hexagon-unknown-elf < %s | FileCheck %s ; In ISelLowering, when folding nodes (or (shl xx, s), (zext y)) @@ -11,17 +12,18 @@ target triple = "hexagon" ; Function Attrs: nofree nosync nounwind memory(readwrite, inaccessiblemem: none) define dso_local void @foo(i64* nocapture noundef %buf, i32 %a, i32 %b) local_unnamed_addr { ; CHECK-LABEL: foo: -; CHECK: // %bb.0: // %entry +; CHECK: .cfi_startproc +; CHECK-NEXT: // %bb.0: // %entry ; CHECK-NEXT: { -; CHECK-NEXT: r[[REG0:[0-9]+]] = addasl(r2,r1,#1) -; CHECK-NEXT: r[[REG2:[0-9]+]] = asl(r1,#1) +; CHECK-NEXT: r2 = addasl(r2,r1,#1) +; CHECK-NEXT: r3 = asl(r1,#1) ; CHECK-NEXT: } ; CHECK-NEXT: { -; CHECK-NEXT: r[[REG1:[0-9]+]] = addasl(r[[REG0]],r1,#1) +; CHECK-NEXT: r2 = addasl(r2,r1,#1) ; CHECK-NEXT: } ; CHECK-NEXT: { ; CHECK-NEXT: jumpr r31 -; CHECK-NEXT: memd(r0+#8) = r[[REG2]]:[[REG1]] +; CHECK-NEXT: memd(r0+#8) = r3:2 ; CHECK-NEXT: } entry: %arrayidx = getelementptr inbounds i64, i64* %buf, i32 1 diff --git a/llvm/test/CodeGen/Hexagon/isel-inttofp-v32i1tov32f32.ll b/llvm/test/CodeGen/Hexagon/isel-inttofp-v32i1tov32f32.ll new file mode 100644 index 0000000..93ca3a2 --- /dev/null +++ b/llvm/test/CodeGen/Hexagon/isel-inttofp-v32i1tov32f32.ll @@ -0,0 +1,42 @@ +; Tests lowering of v32i1 to v32f32 + +; RUN: llc -march=hexagon -mattr=+hvxv79,+hvx-length128b,+hvx-ieee-fp \ +; RUN: -stop-after=hexagon-isel %s -o - | FileCheck %s + +define <32 x float> @uitofp_i1(<32 x i16> %in0, <32 x i16> %in1) #0 { +; CHECK: name: uitofp_i1 +; CHECK: [[R0:%[0-9]+]]:hvxvr = V6_lvsplatw killed %{{[0-9]+}} +; CHECK-NEXT: [[R1:%[0-9]+]]:intregs = A2_tfrsi 1 +; CHECK-NEXT: [[R2:%[0-9]+]]:hvxvr = V6_lvsplatw [[R1]] +; CHECK-NEXT: [[R3:%[0-9]+]]:hvxqr = V6_vandvrt [[R2]], [[R1]] +; CHECK-NEXT: [[R4:%[0-9]+]]:hvxvr = V6_vprefixqw killed [[R3]] +; CHECK-NEXT: [[R5:%[0-9]+]]:hvxvr = V6_vsubw killed [[R4]], [[R2]] +; CHECK-NEXT: [[R6:%[0-9]+]]:hvxvr = V6_vlsrwv killed [[R0]], killed [[R5]] +; CHECK-NEXT: [[R7:%[0-9]+]]:hvxvr = V6_vand killed [[R6]], [[R2]] +; CHECK-NEXT: [[R8:%[0-9]+]]:hvxvr = V6_vconv_sf_w killed [[R7]] +; CHECK-NEXT: hvxvr = V6_vadd_sf_sf [[R8]], [[R8]] + %q1 = icmp eq <32 x i16> %in0, %in1 + %fp0 = uitofp <32 x i1> %q1 to <32 x float> + %out = fadd <32 x float> %fp0, %fp0 + ret <32 x float> %out +} + +define <32 x float> @sitofp_i1(<32 x i16> %in0, <32 x i16> %in1) #0 { +; CHECK: name: sitofp_i1 +; CHECK: [[R0:%[0-9]+]]:hvxvr = V6_lvsplatw killed %{{[0-9]+}} +; CHECK-NEXT: [[R1:%[0-9]+]]:intregs = A2_tfrsi 1 +; CHECK-NEXT: [[R2:%[0-9]+]]:hvxvr = V6_lvsplatw [[R1]] +; CHECK-NEXT: [[R3:%[0-9]+]]:hvxqr = V6_vandvrt [[R2]], [[R1]] +; CHECK-NEXT: [[R4:%[0-9]+]]:hvxvr = V6_vprefixqw killed [[R3]] +; CHECK-NEXT: [[R5:%[0-9]+]]:hvxvr = V6_vsubw killed [[R4]], [[R2]] +; CHECK-NEXT: [[R6:%[0-9]+]]:hvxvr = V6_vlsrwv killed [[R0]], killed [[R5]] +; CHECK-NEXT: [[R7:%[0-9]+]]:hvxvr = V6_vand killed [[R6]], [[R2]] +; CHECK-NEXT: [[R8:%[0-9]+]]:hvxvr = V6_vconv_sf_w killed [[R7]] +; CHECK-NEXT: hvxvr = V6_vadd_sf_sf [[R8]], [[R8]] + %q1 = icmp eq <32 x i16> %in0, %in1 + %fp0 = sitofp <32 x i1> %q1 to <32 x float> + %out = fadd <32 x float> %fp0, %fp0 + ret <32 x float> %out +} + +attributes #0 = { nounwind readnone "target-cpu"="hexagonv79" "target-features"="+hvxv79,+hvx-length128b" } diff --git a/llvm/test/CodeGen/Hexagon/isel-uinttofp-v32i1tov32f32.ll b/llvm/test/CodeGen/Hexagon/isel-uinttofp-v32i1tov32f32.ll deleted file mode 100644 index dfb2bc83..0000000 --- a/llvm/test/CodeGen/Hexagon/isel-uinttofp-v32i1tov32f32.ll +++ /dev/null @@ -1,25 +0,0 @@ -; Tests lowering of v32i1 to v32f32 - -; RUN: llc -march=hexagon -mattr=+hvxv79,+hvx-length128b,+hvx-ieee-fp \ -; RUN: -stop-after=hexagon-isel %s -o - | FileCheck %s - -; CHECK: [[R0:%[0-9]+]]:hvxvr = V6_lvsplatw killed %{{[0-9]+}} -; CHECK-NEXT: [[R1:%[0-9]+]]:intregs = A2_tfrsi 1 -; CHECK-NEXT: [[R2:%[0-9]+]]:hvxvr = V6_lvsplatw [[R1]] -; CHECK-NEXT: [[R3:%[0-9]+]]:hvxqr = V6_vandvrt [[R2]], [[R1]] -; CHECK-NEXT: [[R4:%[0-9]+]]:hvxvr = V6_vprefixqw killed [[R3]] -; CHECK-NEXT: [[R5:%[0-9]+]]:hvxvr = V6_vsubw killed [[R4]], [[R2]] -; CHECK-NEXT: [[R6:%[0-9]+]]:hvxvr = V6_vlsrwv killed [[R0]], killed [[R5]] -; CHECK-NEXT: [[R7:%[0-9]+]]:hvxvr = V6_vand killed [[R6]], [[R2]] -; CHECK-NEXT: [[R8:%[0-9]+]]:hvxvr = V6_vconv_sf_w killed [[R7]] -; CHECK-NEXT: hvxvr = V6_vadd_sf_sf [[R8]], [[R8]] - -define <32 x float> @uitofp_i1(<32 x i16> %in0, <32 x i16> %in1) #0 -{ - %q1 = icmp eq <32 x i16> %in0, %in1 - %fp0 = uitofp <32 x i1> %q1 to <32 x float> - %out = fadd <32 x float> %fp0, %fp0 - ret <32 x float> %out -} - -attributes #0 = { nounwind readnone "target-cpu"="hexagonv79" "target-features"="+hvxv79,+hvx-length128b" } diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/double-intrinsics.ll b/llvm/test/CodeGen/RISCV/GlobalISel/double-intrinsics.ll index 1469d49..4b0acda 100644 --- a/llvm/test/CodeGen/RISCV/GlobalISel/double-intrinsics.ll +++ b/llvm/test/CodeGen/RISCV/GlobalISel/double-intrinsics.ll @@ -1420,3 +1420,61 @@ define double @tanh_f64(double %a) nounwind { %1 = call double @llvm.tanh.f64(double %a) ret double %1 } + +define { double, double } @test_modf_f64(double %a) nounwind { +; RV32IFD-LABEL: test_modf_f64: +; RV32IFD: # %bb.0: +; RV32IFD-NEXT: addi sp, sp, -16 +; RV32IFD-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IFD-NEXT: mv a0, sp +; RV32IFD-NEXT: call modf +; RV32IFD-NEXT: fld fa1, 0(sp) +; RV32IFD-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IFD-NEXT: addi sp, sp, 16 +; RV32IFD-NEXT: ret +; +; RV64IFD-LABEL: test_modf_f64: +; RV64IFD: # %bb.0: +; RV64IFD-NEXT: addi sp, sp, -16 +; RV64IFD-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IFD-NEXT: mv a0, sp +; RV64IFD-NEXT: call modf +; RV64IFD-NEXT: fld fa1, 0(sp) +; RV64IFD-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IFD-NEXT: addi sp, sp, 16 +; RV64IFD-NEXT: ret +; +; RV32I-LABEL: test_modf_f64: +; RV32I: # %bb.0: +; RV32I-NEXT: addi sp, sp, -16 +; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32I-NEXT: sw s0, 8(sp) # 4-byte Folded Spill +; RV32I-NEXT: mv s0, a0 +; RV32I-NEXT: mv a0, a1 +; RV32I-NEXT: mv a1, a2 +; RV32I-NEXT: mv a2, sp +; RV32I-NEXT: call modf +; RV32I-NEXT: lw a2, 0(sp) +; RV32I-NEXT: lw a3, 4(sp) +; RV32I-NEXT: sw a0, 0(s0) +; RV32I-NEXT: sw a1, 4(s0) +; RV32I-NEXT: sw a2, 8(s0) +; RV32I-NEXT: sw a3, 12(s0) +; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32I-NEXT: lw s0, 8(sp) # 4-byte Folded Reload +; RV32I-NEXT: addi sp, sp, 16 +; RV32I-NEXT: ret +; +; RV64I-LABEL: test_modf_f64: +; RV64I: # %bb.0: +; RV64I-NEXT: addi sp, sp, -16 +; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64I-NEXT: mv a1, sp +; RV64I-NEXT: call modf +; RV64I-NEXT: ld a1, 0(sp) +; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64I-NEXT: addi sp, sp, 16 +; RV64I-NEXT: ret + %result = call { double, double } @llvm.modf.f64(double %a) + ret { double, double } %result +} diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/float-intrinsics.ll b/llvm/test/CodeGen/RISCV/GlobalISel/float-intrinsics.ll index 23f660b..01d9ceb 100644 --- a/llvm/test/CodeGen/RISCV/GlobalISel/float-intrinsics.ll +++ b/llvm/test/CodeGen/RISCV/GlobalISel/float-intrinsics.ll @@ -2118,3 +2118,62 @@ define float @tanh_f32(float %a) nounwind { %1 = call float @llvm.tanh.f32(float %a) ret float %1 } + +define { float, float } @test_modf_f32(float %a) nounwind { +; RV32IF-LABEL: test_modf_f32: +; RV32IF: # %bb.0: +; RV32IF-NEXT: addi sp, sp, -16 +; RV32IF-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IF-NEXT: addi a0, sp, 8 +; RV32IF-NEXT: call modff +; RV32IF-NEXT: flw fa1, 8(sp) +; RV32IF-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IF-NEXT: addi sp, sp, 16 +; RV32IF-NEXT: ret +; +; RV64IF-LABEL: test_modf_f32: +; RV64IF: # %bb.0: +; RV64IF-NEXT: addi sp, sp, -16 +; RV64IF-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IF-NEXT: addi a0, sp, 4 +; RV64IF-NEXT: call modff +; RV64IF-NEXT: flw fa1, 4(sp) +; RV64IF-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IF-NEXT: addi sp, sp, 16 +; RV64IF-NEXT: ret +; +; RV64IFD-LABEL: test_modf_f32: +; RV64IFD: # %bb.0: +; RV64IFD-NEXT: addi sp, sp, -16 +; RV64IFD-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IFD-NEXT: addi a0, sp, 4 +; RV64IFD-NEXT: call modff +; RV64IFD-NEXT: flw fa1, 4(sp) +; RV64IFD-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IFD-NEXT: addi sp, sp, 16 +; RV64IFD-NEXT: ret +; +; RV32I-LABEL: test_modf_f32: +; RV32I: # %bb.0: +; RV32I-NEXT: addi sp, sp, -16 +; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32I-NEXT: addi a1, sp, 8 +; RV32I-NEXT: call modff +; RV32I-NEXT: lw a1, 8(sp) +; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32I-NEXT: addi sp, sp, 16 +; RV32I-NEXT: ret +; +; RV64I-LABEL: test_modf_f32: +; RV64I: # %bb.0: +; RV64I-NEXT: addi sp, sp, -16 +; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64I-NEXT: addi a1, sp, 4 +; RV64I-NEXT: call modff +; RV64I-NEXT: lw a1, 4(sp) +; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64I-NEXT: addi sp, sp, 16 +; RV64I-NEXT: ret + %result = call { float, float } @llvm.modf.f32(float %a) + ret { float, float } %result +} diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll b/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll index ae9b6cc..e7a3f79 100644 --- a/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll +++ b/llvm/test/CodeGen/RISCV/GlobalISel/fp128.ll @@ -911,3 +911,28 @@ define fp128 @tanh(fp128 %a) nounwind { %1 = call fp128 @llvm.tanh.f128(fp128 %a) ret fp128 %1 } + +define { fp128, fp128 } @modf(fp128 %a) nounwind { +; CHECK-LABEL: modf: +; CHECK: # %bb.0: +; CHECK-NEXT: addi sp, sp, -32 +; CHECK-NEXT: sd ra, 24(sp) # 8-byte Folded Spill +; CHECK-NEXT: sd s0, 16(sp) # 8-byte Folded Spill +; CHECK-NEXT: mv s0, a0 +; CHECK-NEXT: mv a0, a1 +; CHECK-NEXT: mv a1, a2 +; CHECK-NEXT: mv a2, sp +; CHECK-NEXT: call modfl +; CHECK-NEXT: ld a2, 0(sp) +; CHECK-NEXT: ld a3, 8(sp) +; CHECK-NEXT: sd a0, 0(s0) +; CHECK-NEXT: sd a1, 8(s0) +; CHECK-NEXT: sd a2, 16(s0) +; CHECK-NEXT: sd a3, 24(s0) +; CHECK-NEXT: ld ra, 24(sp) # 8-byte Folded Reload +; CHECK-NEXT: ld s0, 16(sp) # 8-byte Folded Reload +; CHECK-NEXT: addi sp, sp, 32 +; CHECK-NEXT: ret + %result = call { fp128, fp128 } @llvm.modf.f128(fp128 %a) + ret { fp128, fp128 } %result +} diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/legalizer-info-validation.mir b/llvm/test/CodeGen/RISCV/GlobalISel/legalizer-info-validation.mir index f1d17f9f..1361d92 100644 --- a/llvm/test/CodeGen/RISCV/GlobalISel/legalizer-info-validation.mir +++ b/llvm/test/CodeGen/RISCV/GlobalISel/legalizer-info-validation.mir @@ -506,8 +506,9 @@ # DEBUG-NEXT: .. the first uncovered type index: 1, OK # DEBUG-NEXT: .. the first uncovered imm index: 0, OK # DEBUG-NEXT: G_FMODF (opcode {{[0-9]+}}): 1 type index, 0 imm indices -# DEBUG-NEXT: .. type index coverage check SKIPPED: no rules defined -# DEBUG-NEXT: .. imm index coverage check SKIPPED: no rules defined +# DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}} +# DEBUG-NEXT: .. the first uncovered type index: 1, OK +# DEBUG-NEXT: .. the first uncovered imm index: 0, OK # DEBUG-NEXT: G_FPOW (opcode {{[0-9]+}}): 1 type index, 0 imm indices # DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}} # DEBUG-NEXT: .. the first uncovered type index: 1, OK diff --git a/llvm/test/CodeGen/RISCV/double-intrinsics.ll b/llvm/test/CodeGen/RISCV/double-intrinsics.ll index caeb6e6..aaa08b5 100644 --- a/llvm/test/CodeGen/RISCV/double-intrinsics.ll +++ b/llvm/test/CodeGen/RISCV/double-intrinsics.ll @@ -2109,3 +2109,85 @@ define double @tanh_f64(double %a) nounwind { %1 = call double @llvm.tanh.f64(double %a) ret double %1 } + +define { double, double } @test_modf_f64(double %a) nounwind { +; RV32IFD-LABEL: test_modf_f64: +; RV32IFD: # %bb.0: +; RV32IFD-NEXT: addi sp, sp, -16 +; RV32IFD-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IFD-NEXT: mv a0, sp +; RV32IFD-NEXT: call modf +; RV32IFD-NEXT: fld fa1, 0(sp) +; RV32IFD-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IFD-NEXT: addi sp, sp, 16 +; RV32IFD-NEXT: ret +; +; RV64IFD-LABEL: test_modf_f64: +; RV64IFD: # %bb.0: +; RV64IFD-NEXT: addi sp, sp, -16 +; RV64IFD-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IFD-NEXT: mv a0, sp +; RV64IFD-NEXT: call modf +; RV64IFD-NEXT: fld fa1, 0(sp) +; RV64IFD-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IFD-NEXT: addi sp, sp, 16 +; RV64IFD-NEXT: ret +; +; RV32IZFINXZDINX-LABEL: test_modf_f64: +; RV32IZFINXZDINX: # %bb.0: +; RV32IZFINXZDINX-NEXT: addi sp, sp, -16 +; RV32IZFINXZDINX-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IZFINXZDINX-NEXT: mv a2, sp +; RV32IZFINXZDINX-NEXT: call modf +; RV32IZFINXZDINX-NEXT: lw a2, 0(sp) +; RV32IZFINXZDINX-NEXT: lw a3, 4(sp) +; RV32IZFINXZDINX-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IZFINXZDINX-NEXT: addi sp, sp, 16 +; RV32IZFINXZDINX-NEXT: ret +; +; RV64IZFINXZDINX-LABEL: test_modf_f64: +; RV64IZFINXZDINX: # %bb.0: +; RV64IZFINXZDINX-NEXT: addi sp, sp, -16 +; RV64IZFINXZDINX-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IZFINXZDINX-NEXT: mv a1, sp +; RV64IZFINXZDINX-NEXT: call modf +; RV64IZFINXZDINX-NEXT: ld a1, 0(sp) +; RV64IZFINXZDINX-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IZFINXZDINX-NEXT: addi sp, sp, 16 +; RV64IZFINXZDINX-NEXT: ret +; +; RV32I-LABEL: test_modf_f64: +; RV32I: # %bb.0: +; RV32I-NEXT: addi sp, sp, -16 +; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32I-NEXT: sw s0, 8(sp) # 4-byte Folded Spill +; RV32I-NEXT: mv a3, a2 +; RV32I-NEXT: mv s0, a0 +; RV32I-NEXT: mv a2, sp +; RV32I-NEXT: mv a0, a1 +; RV32I-NEXT: mv a1, a3 +; RV32I-NEXT: call modf +; RV32I-NEXT: lw a2, 0(sp) +; RV32I-NEXT: lw a3, 4(sp) +; RV32I-NEXT: sw a0, 0(s0) +; RV32I-NEXT: sw a1, 4(s0) +; RV32I-NEXT: sw a2, 8(s0) +; RV32I-NEXT: sw a3, 12(s0) +; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32I-NEXT: lw s0, 8(sp) # 4-byte Folded Reload +; RV32I-NEXT: addi sp, sp, 16 +; RV32I-NEXT: ret +; +; RV64I-LABEL: test_modf_f64: +; RV64I: # %bb.0: +; RV64I-NEXT: addi sp, sp, -16 +; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64I-NEXT: mv a1, sp +; RV64I-NEXT: call modf +; RV64I-NEXT: ld a1, 0(sp) +; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64I-NEXT: addi sp, sp, 16 +; RV64I-NEXT: ret + %result = call { double, double } @llvm.modf.f64(double %a) + ret { double, double } %result +} diff --git a/llvm/test/CodeGen/RISCV/float-intrinsics.ll b/llvm/test/CodeGen/RISCV/float-intrinsics.ll index b1230ae..5f673ac 100644 --- a/llvm/test/CodeGen/RISCV/float-intrinsics.ll +++ b/llvm/test/CodeGen/RISCV/float-intrinsics.ll @@ -3050,3 +3050,84 @@ define float @tanh_f32(float %a) nounwind { %1 = call float @llvm.tanh.f32(float %a) ret float %1 } + +define { float, float } @test_modf_f32(float %a) nounwind { +; RV32IF-LABEL: test_modf_f32: +; RV32IF: # %bb.0: +; RV32IF-NEXT: addi sp, sp, -16 +; RV32IF-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IF-NEXT: addi a0, sp, 8 +; RV32IF-NEXT: call modff +; RV32IF-NEXT: flw fa1, 8(sp) +; RV32IF-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IF-NEXT: addi sp, sp, 16 +; RV32IF-NEXT: ret +; +; RV32IZFINX-LABEL: test_modf_f32: +; RV32IZFINX: # %bb.0: +; RV32IZFINX-NEXT: addi sp, sp, -16 +; RV32IZFINX-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IZFINX-NEXT: addi a1, sp, 8 +; RV32IZFINX-NEXT: call modff +; RV32IZFINX-NEXT: lw a1, 8(sp) +; RV32IZFINX-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IZFINX-NEXT: addi sp, sp, 16 +; RV32IZFINX-NEXT: ret +; +; RV64IF-LABEL: test_modf_f32: +; RV64IF: # %bb.0: +; RV64IF-NEXT: addi sp, sp, -16 +; RV64IF-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IF-NEXT: addi a0, sp, 4 +; RV64IF-NEXT: call modff +; RV64IF-NEXT: flw fa1, 4(sp) +; RV64IF-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IF-NEXT: addi sp, sp, 16 +; RV64IF-NEXT: ret +; +; RV64IZFINX-LABEL: test_modf_f32: +; RV64IZFINX: # %bb.0: +; RV64IZFINX-NEXT: addi sp, sp, -16 +; RV64IZFINX-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IZFINX-NEXT: addi a1, sp, 4 +; RV64IZFINX-NEXT: call modff +; RV64IZFINX-NEXT: lw a1, 4(sp) +; RV64IZFINX-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IZFINX-NEXT: addi sp, sp, 16 +; RV64IZFINX-NEXT: ret +; +; RV64IFD-LABEL: test_modf_f32: +; RV64IFD: # %bb.0: +; RV64IFD-NEXT: addi sp, sp, -16 +; RV64IFD-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IFD-NEXT: addi a0, sp, 4 +; RV64IFD-NEXT: call modff +; RV64IFD-NEXT: flw fa1, 4(sp) +; RV64IFD-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IFD-NEXT: addi sp, sp, 16 +; RV64IFD-NEXT: ret +; +; RV32I-LABEL: test_modf_f32: +; RV32I: # %bb.0: +; RV32I-NEXT: addi sp, sp, -16 +; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32I-NEXT: addi a1, sp, 8 +; RV32I-NEXT: call modff +; RV32I-NEXT: lw a1, 8(sp) +; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32I-NEXT: addi sp, sp, 16 +; RV32I-NEXT: ret +; +; RV64I-LABEL: test_modf_f32: +; RV64I: # %bb.0: +; RV64I-NEXT: addi sp, sp, -16 +; RV64I-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64I-NEXT: addi a1, sp, 4 +; RV64I-NEXT: call modff +; RV64I-NEXT: lw a1, 4(sp) +; RV64I-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64I-NEXT: addi sp, sp, 16 +; RV64I-NEXT: ret + %result = call { float, float } @llvm.modf.f32(float %a) + ret { float, float } %result +} diff --git a/llvm/test/CodeGen/RISCV/fp128.ll b/llvm/test/CodeGen/RISCV/fp128.ll index a8e26f7..704cb425 100644 --- a/llvm/test/CodeGen/RISCV/fp128.ll +++ b/llvm/test/CodeGen/RISCV/fp128.ll @@ -189,3 +189,46 @@ define fp128 @fminimum(fp128 %x, fp128 %y) { %a = call fp128 @llvm.minimum.fp128(fp128 %x, fp128 %y) ret fp128 %a } + +define { fp128, fp128 } @modf(fp128 %a) nounwind { +; RV32I-LABEL: modf: +; RV32I: # %bb.0: +; RV32I-NEXT: addi sp, sp, -64 +; RV32I-NEXT: sw ra, 60(sp) # 4-byte Folded Spill +; RV32I-NEXT: sw s0, 56(sp) # 4-byte Folded Spill +; RV32I-NEXT: lw a3, 0(a1) +; RV32I-NEXT: lw a4, 4(a1) +; RV32I-NEXT: lw a5, 8(a1) +; RV32I-NEXT: lw a6, 12(a1) +; RV32I-NEXT: mv s0, a0 +; RV32I-NEXT: addi a0, sp, 24 +; RV32I-NEXT: addi a1, sp, 8 +; RV32I-NEXT: addi a2, sp, 40 +; RV32I-NEXT: sw a3, 8(sp) +; RV32I-NEXT: sw a4, 12(sp) +; RV32I-NEXT: sw a5, 16(sp) +; RV32I-NEXT: sw a6, 20(sp) +; RV32I-NEXT: call modfl +; RV32I-NEXT: lw a0, 24(sp) +; RV32I-NEXT: lw a1, 28(sp) +; RV32I-NEXT: lw a2, 32(sp) +; RV32I-NEXT: lw a3, 36(sp) +; RV32I-NEXT: lw a4, 40(sp) +; RV32I-NEXT: lw a5, 44(sp) +; RV32I-NEXT: lw a6, 48(sp) +; RV32I-NEXT: lw a7, 52(sp) +; RV32I-NEXT: sw a4, 16(s0) +; RV32I-NEXT: sw a5, 20(s0) +; RV32I-NEXT: sw a6, 24(s0) +; RV32I-NEXT: sw a7, 28(s0) +; RV32I-NEXT: sw a0, 0(s0) +; RV32I-NEXT: sw a1, 4(s0) +; RV32I-NEXT: sw a2, 8(s0) +; RV32I-NEXT: sw a3, 12(s0) +; RV32I-NEXT: lw ra, 60(sp) # 4-byte Folded Reload +; RV32I-NEXT: lw s0, 56(sp) # 4-byte Folded Reload +; RV32I-NEXT: addi sp, sp, 64 +; RV32I-NEXT: ret + %result = call { fp128, fp128 } @llvm.modf.f128(fp128 %a) + ret { fp128, fp128 } %result +} diff --git a/llvm/test/CodeGen/RISCV/half-intrinsics.ll b/llvm/test/CodeGen/RISCV/half-intrinsics.ll index e16d788..847054d 100644 --- a/llvm/test/CodeGen/RISCV/half-intrinsics.ll +++ b/llvm/test/CodeGen/RISCV/half-intrinsics.ll @@ -4417,3 +4417,161 @@ define half @tanh_f16(half %a) nounwind { %1 = call half @llvm.tanh.f16(half %a) ret half %1 } + +define { half, half } @test_modf_f16(half %a) nounwind { +; RV32IZFH-LABEL: test_modf_f16: +; RV32IZFH: # %bb.0: +; RV32IZFH-NEXT: addi sp, sp, -16 +; RV32IZFH-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IZFH-NEXT: fcvt.s.h fa0, fa0 +; RV32IZFH-NEXT: addi a0, sp, 8 +; RV32IZFH-NEXT: call modff +; RV32IZFH-NEXT: flw fa5, 8(sp) +; RV32IZFH-NEXT: fcvt.h.s fa0, fa0 +; RV32IZFH-NEXT: fcvt.h.s fa1, fa5 +; RV32IZFH-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IZFH-NEXT: addi sp, sp, 16 +; RV32IZFH-NEXT: ret +; +; RV64IZFH-LABEL: test_modf_f16: +; RV64IZFH: # %bb.0: +; RV64IZFH-NEXT: addi sp, sp, -16 +; RV64IZFH-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IZFH-NEXT: fcvt.s.h fa0, fa0 +; RV64IZFH-NEXT: addi a0, sp, 4 +; RV64IZFH-NEXT: call modff +; RV64IZFH-NEXT: flw fa5, 4(sp) +; RV64IZFH-NEXT: fcvt.h.s fa0, fa0 +; RV64IZFH-NEXT: fcvt.h.s fa1, fa5 +; RV64IZFH-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IZFH-NEXT: addi sp, sp, 16 +; RV64IZFH-NEXT: ret +; +; RV32IZHINX-LABEL: test_modf_f16: +; RV32IZHINX: # %bb.0: +; RV32IZHINX-NEXT: addi sp, sp, -16 +; RV32IZHINX-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IZHINX-NEXT: fcvt.s.h a0, a0 +; RV32IZHINX-NEXT: addi a1, sp, 8 +; RV32IZHINX-NEXT: call modff +; RV32IZHINX-NEXT: lw a1, 8(sp) +; RV32IZHINX-NEXT: fcvt.h.s a0, a0 +; RV32IZHINX-NEXT: fcvt.h.s a1, a1 +; RV32IZHINX-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IZHINX-NEXT: addi sp, sp, 16 +; RV32IZHINX-NEXT: ret +; +; RV64IZHINX-LABEL: test_modf_f16: +; RV64IZHINX: # %bb.0: +; RV64IZHINX-NEXT: addi sp, sp, -16 +; RV64IZHINX-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IZHINX-NEXT: fcvt.s.h a0, a0 +; RV64IZHINX-NEXT: addi a1, sp, 4 +; RV64IZHINX-NEXT: call modff +; RV64IZHINX-NEXT: lw a1, 4(sp) +; RV64IZHINX-NEXT: fcvt.h.s a0, a0 +; RV64IZHINX-NEXT: fcvt.h.s a1, a1 +; RV64IZHINX-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IZHINX-NEXT: addi sp, sp, 16 +; RV64IZHINX-NEXT: ret +; +; RV32I-LABEL: test_modf_f16: +; RV32I: # %bb.0: +; RV32I-NEXT: addi sp, sp, -16 +; RV32I-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32I-NEXT: sw s0, 8(sp) # 4-byte Folded Spill +; RV32I-NEXT: slli a0, a0, 16 +; RV32I-NEXT: srli a0, a0, 16 +; RV32I-NEXT: call __extendhfsf2 +; RV32I-NEXT: addi a1, sp, 4 +; RV32I-NEXT: call modff +; RV32I-NEXT: call __truncsfhf2 +; RV32I-NEXT: mv s0, a0 +; RV32I-NEXT: lw a0, 4(sp) +; RV32I-NEXT: call __truncsfhf2 +; RV32I-NEXT: mv a1, a0 +; RV32I-NEXT: mv a0, s0 +; RV32I-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32I-NEXT: lw s0, 8(sp) # 4-byte Folded Reload +; RV32I-NEXT: addi sp, sp, 16 +; RV32I-NEXT: ret +; +; RV64I-LABEL: test_modf_f16: +; RV64I: # %bb.0: +; RV64I-NEXT: addi sp, sp, -32 +; RV64I-NEXT: sd ra, 24(sp) # 8-byte Folded Spill +; RV64I-NEXT: sd s0, 16(sp) # 8-byte Folded Spill +; RV64I-NEXT: slli a0, a0, 48 +; RV64I-NEXT: srli a0, a0, 48 +; RV64I-NEXT: call __extendhfsf2 +; RV64I-NEXT: addi a1, sp, 12 +; RV64I-NEXT: call modff +; RV64I-NEXT: call __truncsfhf2 +; RV64I-NEXT: mv s0, a0 +; RV64I-NEXT: lw a0, 12(sp) +; RV64I-NEXT: call __truncsfhf2 +; RV64I-NEXT: mv a1, a0 +; RV64I-NEXT: mv a0, s0 +; RV64I-NEXT: ld ra, 24(sp) # 8-byte Folded Reload +; RV64I-NEXT: ld s0, 16(sp) # 8-byte Folded Reload +; RV64I-NEXT: addi sp, sp, 32 +; RV64I-NEXT: ret +; +; RV32IZFHMIN-LABEL: test_modf_f16: +; RV32IZFHMIN: # %bb.0: +; RV32IZFHMIN-NEXT: addi sp, sp, -16 +; RV32IZFHMIN-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IZFHMIN-NEXT: fcvt.s.h fa0, fa0 +; RV32IZFHMIN-NEXT: addi a0, sp, 8 +; RV32IZFHMIN-NEXT: call modff +; RV32IZFHMIN-NEXT: flw fa5, 8(sp) +; RV32IZFHMIN-NEXT: fcvt.h.s fa0, fa0 +; RV32IZFHMIN-NEXT: fcvt.h.s fa1, fa5 +; RV32IZFHMIN-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IZFHMIN-NEXT: addi sp, sp, 16 +; RV32IZFHMIN-NEXT: ret +; +; RV64IZFHMIN-LABEL: test_modf_f16: +; RV64IZFHMIN: # %bb.0: +; RV64IZFHMIN-NEXT: addi sp, sp, -16 +; RV64IZFHMIN-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IZFHMIN-NEXT: fcvt.s.h fa0, fa0 +; RV64IZFHMIN-NEXT: addi a0, sp, 4 +; RV64IZFHMIN-NEXT: call modff +; RV64IZFHMIN-NEXT: flw fa5, 4(sp) +; RV64IZFHMIN-NEXT: fcvt.h.s fa0, fa0 +; RV64IZFHMIN-NEXT: fcvt.h.s fa1, fa5 +; RV64IZFHMIN-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IZFHMIN-NEXT: addi sp, sp, 16 +; RV64IZFHMIN-NEXT: ret +; +; RV32IZHINXMIN-LABEL: test_modf_f16: +; RV32IZHINXMIN: # %bb.0: +; RV32IZHINXMIN-NEXT: addi sp, sp, -16 +; RV32IZHINXMIN-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32IZHINXMIN-NEXT: fcvt.s.h a0, a0 +; RV32IZHINXMIN-NEXT: addi a1, sp, 8 +; RV32IZHINXMIN-NEXT: call modff +; RV32IZHINXMIN-NEXT: lw a1, 8(sp) +; RV32IZHINXMIN-NEXT: fcvt.h.s a0, a0 +; RV32IZHINXMIN-NEXT: fcvt.h.s a1, a1 +; RV32IZHINXMIN-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32IZHINXMIN-NEXT: addi sp, sp, 16 +; RV32IZHINXMIN-NEXT: ret +; +; RV64IZHINXMIN-LABEL: test_modf_f16: +; RV64IZHINXMIN: # %bb.0: +; RV64IZHINXMIN-NEXT: addi sp, sp, -16 +; RV64IZHINXMIN-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64IZHINXMIN-NEXT: fcvt.s.h a0, a0 +; RV64IZHINXMIN-NEXT: addi a1, sp, 4 +; RV64IZHINXMIN-NEXT: call modff +; RV64IZHINXMIN-NEXT: lw a1, 4(sp) +; RV64IZHINXMIN-NEXT: fcvt.h.s a0, a0 +; RV64IZHINXMIN-NEXT: fcvt.h.s a1, a1 +; RV64IZHINXMIN-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64IZHINXMIN-NEXT: addi sp, sp, 16 +; RV64IZHINXMIN-NEXT: ret + %result = call { half, half } @llvm.modf.f16(half %a) + ret { half, half } %result +} diff --git a/llvm/test/CodeGen/SPARC/2011-01-19-DelaySlot.ll b/llvm/test/CodeGen/SPARC/2011-01-19-DelaySlot.ll index 9ccd4f1..767ef7e 100644 --- a/llvm/test/CodeGen/SPARC/2011-01-19-DelaySlot.ll +++ b/llvm/test/CodeGen/SPARC/2011-01-19-DelaySlot.ll @@ -184,4 +184,29 @@ entry: ret i32 %2 } +define i32 @test_generic_inst(i32 %arg) #0 { +;CHECK-LABEL: test_generic_inst: +;CHECK: ! fake_use: {{.*}} +;CHECK: bne {{.*}} +;CHECK-NEXT: nop + %bar1 = call i32 @bar(i32 %arg) + %even = and i32 %bar1, 1 + %cmp = icmp eq i32 %even, 0 + ; This shouldn't get reordered into a delay slot + call void (...) @llvm.fake.use(i32 %arg) + br i1 %cmp, label %true, label %false +true: + %bar2 = call i32 @bar(i32 %bar1) + br label %cont + +false: + %inc = add nsw i32 %bar1, 1 + br label %cont + +cont: + %ret = phi i32 [ %bar2, %true ], [ %inc, %false ] + ret i32 %ret +} + +declare void @llvm.fake.use(...) attributes #0 = { nounwind "disable-tail-calls"="true" } diff --git a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll index dec829f..44cf4e8 100644 --- a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll +++ b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll @@ -911,7 +911,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 16(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: paddb (%rdx), %xmm0 ; SSE2-NEXT: movdqa %xmm0, (%rcx) @@ -1898,7 +1898,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 32(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: movdqa %xmm0, %xmm3 ; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3] @@ -4155,7 +4155,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 48(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1] ; SSE2-NEXT: paddb (%rdx), %xmm2 diff --git a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll index 3d4cddb..89b5c33 100644 --- a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll +++ b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll @@ -769,7 +769,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-LABEL: vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] ; SSE2-NEXT: paddb (%rsi), %xmm1 ; SSE2-NEXT: movdqa %xmm1, (%rdx) @@ -1522,7 +1522,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-LABEL: vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: movdqa %xmm1, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] @@ -3335,7 +3335,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2: # %bb.0: ; SSE2-NEXT: movdqa (%rdi), %xmm0 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1] ; SSE2-NEXT: paddb (%rsi), %xmm2 diff --git a/llvm/test/CodeGen/X86/avx2-vector-shifts.ll b/llvm/test/CodeGen/X86/avx2-vector-shifts.ll index 983c69d..95c2eda 100644 --- a/llvm/test/CodeGen/X86/avx2-vector-shifts.ll +++ b/llvm/test/CodeGen/X86/avx2-vector-shifts.ll @@ -441,10 +441,10 @@ define <32 x i8> @ashr_32i8(<32 x i8> %r, <32 x i8> %a) nounwind { ; CHECK-NEXT: vpsraw $4, %ymm3, %ymm4 ; CHECK-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; CHECK-NEXT: vpsraw $2, %ymm3, %ymm4 -; CHECK-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; CHECK-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; CHECK-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; CHECK-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; CHECK-NEXT: vpsraw $1, %ymm3, %ymm4 -; CHECK-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; CHECK-NEXT: vpsllw $2, %ymm2, %ymm2 ; CHECK-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; CHECK-NEXT: vpsrlw $8, %ymm2, %ymm2 ; CHECK-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -452,10 +452,10 @@ define <32 x i8> @ashr_32i8(<32 x i8> %r, <32 x i8> %a) nounwind { ; CHECK-NEXT: vpsraw $4, %ymm0, %ymm3 ; CHECK-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; CHECK-NEXT: vpsraw $2, %ymm0, %ymm3 -; CHECK-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; CHECK-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; CHECK-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; CHECK-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; CHECK-NEXT: vpsraw $1, %ymm0, %ymm3 -; CHECK-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; CHECK-NEXT: vpsllw $2, %ymm1, %ymm1 ; CHECK-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; CHECK-NEXT: vpsrlw $8, %ymm0, %ymm0 ; CHECK-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 diff --git a/llvm/test/CodeGen/X86/gfni-shifts.ll b/llvm/test/CodeGen/X86/gfni-shifts.ll index cd16651..feac3dc 100644 --- a/llvm/test/CodeGen/X86/gfni-shifts.ll +++ b/llvm/test/CodeGen/X86/gfni-shifts.ll @@ -166,10 +166,10 @@ define <16 x i8> @var_ashr_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind { ; GFNIAVX1OR2-NEXT: vpsraw $4, %xmm3, %xmm4 ; GFNIAVX1OR2-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 ; GFNIAVX1OR2-NEXT: vpsraw $2, %xmm3, %xmm4 -; GFNIAVX1OR2-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; GFNIAVX1OR2-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 +; GFNIAVX1OR2-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; GFNIAVX1OR2-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; GFNIAVX1OR2-NEXT: vpsraw $1, %xmm3, %xmm4 -; GFNIAVX1OR2-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; GFNIAVX1OR2-NEXT: vpsllw $2, %xmm2, %xmm2 ; GFNIAVX1OR2-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; GFNIAVX1OR2-NEXT: vpsrlw $8, %xmm2, %xmm2 ; GFNIAVX1OR2-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -177,10 +177,10 @@ define <16 x i8> @var_ashr_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind { ; GFNIAVX1OR2-NEXT: vpsraw $4, %xmm0, %xmm3 ; GFNIAVX1OR2-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; GFNIAVX1OR2-NEXT: vpsraw $2, %xmm0, %xmm3 -; GFNIAVX1OR2-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; GFNIAVX1OR2-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 +; GFNIAVX1OR2-NEXT: vpaddw %xmm1, %xmm1, %xmm4 +; GFNIAVX1OR2-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 ; GFNIAVX1OR2-NEXT: vpsraw $1, %xmm0, %xmm3 -; GFNIAVX1OR2-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; GFNIAVX1OR2-NEXT: vpsllw $2, %xmm1, %xmm1 ; GFNIAVX1OR2-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; GFNIAVX1OR2-NEXT: vpsrlw $8, %xmm0, %xmm0 ; GFNIAVX1OR2-NEXT: vpackuswb %xmm2, %xmm0, %xmm0 @@ -896,10 +896,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm5, %xmm6 ; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpsraw $2, %xmm5, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm5 +; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm7 +; GFNIAVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpsraw $1, %xmm5, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 +; GFNIAVX1-NEXT: vpsllw $2, %xmm3, %xmm3 ; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm3 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm3, %xmm3 ; GFNIAVX1-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -907,10 +907,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm4, %xmm5 ; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpsraw $2, %xmm4, %xmm5 -; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 +; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm6 +; GFNIAVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpsraw $1, %xmm4, %xmm5 -; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; GFNIAVX1-NEXT: vpsllw $2, %xmm2, %xmm2 ; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm2 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm2, %xmm2 ; GFNIAVX1-NEXT: vpackuswb %xmm3, %xmm2, %xmm2 @@ -920,10 +920,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm4, %xmm5 ; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpsraw $2, %xmm4, %xmm5 -; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm4 +; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm6 +; GFNIAVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpsraw $1, %xmm4, %xmm5 -; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 +; GFNIAVX1-NEXT: vpsllw $2, %xmm3, %xmm3 ; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm3 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm3, %xmm3 ; GFNIAVX1-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -931,10 +931,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm0, %xmm4 ; GFNIAVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpsraw $2, %xmm0, %xmm4 -; GFNIAVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; GFNIAVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 +; GFNIAVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm5 +; GFNIAVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpsraw $1, %xmm0, %xmm4 -; GFNIAVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; GFNIAVX1-NEXT: vpsllw $2, %xmm1, %xmm1 ; GFNIAVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpackuswb %xmm3, %xmm0, %xmm0 @@ -949,10 +949,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX2-NEXT: vpsraw $4, %ymm3, %ymm4 ; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; GFNIAVX2-NEXT: vpsraw $2, %ymm3, %ymm4 -; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; GFNIAVX2-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; GFNIAVX2-NEXT: vpsraw $1, %ymm3, %ymm4 -; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; GFNIAVX2-NEXT: vpsllw $2, %ymm2, %ymm2 ; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; GFNIAVX2-NEXT: vpsrlw $8, %ymm2, %ymm2 ; GFNIAVX2-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -960,10 +960,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX2-NEXT: vpsraw $4, %ymm0, %ymm3 ; GFNIAVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpsraw $2, %ymm0, %ymm3 -; GFNIAVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; GFNIAVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; GFNIAVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; GFNIAVX2-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpsraw $1, %ymm0, %ymm3 -; GFNIAVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; GFNIAVX2-NEXT: vpsllw $2, %ymm1, %ymm1 ; GFNIAVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpsrlw $8, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 @@ -977,10 +977,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX512VL-NEXT: vpsraw $4, %ymm3, %ymm4 ; GFNIAVX512VL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; GFNIAVX512VL-NEXT: vpsraw $2, %ymm3, %ymm4 -; GFNIAVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; GFNIAVX512VL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; GFNIAVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; GFNIAVX512VL-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; GFNIAVX512VL-NEXT: vpsraw $1, %ymm3, %ymm4 -; GFNIAVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; GFNIAVX512VL-NEXT: vpsllw $2, %ymm2, %ymm2 ; GFNIAVX512VL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; GFNIAVX512VL-NEXT: vpsrlw $8, %ymm2, %ymm2 ; GFNIAVX512VL-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -988,10 +988,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; GFNIAVX512VL-NEXT: vpsraw $4, %ymm0, %ymm3 ; GFNIAVX512VL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpsraw $2, %ymm0, %ymm3 -; GFNIAVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; GFNIAVX512VL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; GFNIAVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; GFNIAVX512VL-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpsraw $1, %ymm0, %ymm3 -; GFNIAVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; GFNIAVX512VL-NEXT: vpsllw $2, %ymm1, %ymm1 ; GFNIAVX512VL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpsrlw $8, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 @@ -2027,10 +2027,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm7, %xmm8 ; GFNIAVX1-NEXT: vpblendvb %xmm5, %xmm8, %xmm7, %xmm7 ; GFNIAVX1-NEXT: vpsraw $2, %xmm7, %xmm8 -; GFNIAVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm5 -; GFNIAVX1-NEXT: vpblendvb %xmm5, %xmm8, %xmm7, %xmm7 +; GFNIAVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm9 +; GFNIAVX1-NEXT: vpblendvb %xmm9, %xmm8, %xmm7, %xmm7 ; GFNIAVX1-NEXT: vpsraw $1, %xmm7, %xmm8 -; GFNIAVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm5 +; GFNIAVX1-NEXT: vpsllw $2, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpblendvb %xmm5, %xmm8, %xmm7, %xmm5 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpunpcklbw {{.*#+}} xmm4 = xmm4[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -2038,10 +2038,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm6, %xmm7 ; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm7, %xmm6, %xmm6 ; GFNIAVX1-NEXT: vpsraw $2, %xmm6, %xmm7 -; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm4 -; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm7, %xmm6, %xmm6 +; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm8 +; GFNIAVX1-NEXT: vpblendvb %xmm8, %xmm7, %xmm6, %xmm6 ; GFNIAVX1-NEXT: vpsraw $1, %xmm6, %xmm7 -; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm4 +; GFNIAVX1-NEXT: vpsllw $2, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm7, %xmm6, %xmm4 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpackuswb %xmm5, %xmm4, %xmm4 @@ -2051,10 +2051,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm6, %xmm7 ; GFNIAVX1-NEXT: vpblendvb %xmm5, %xmm7, %xmm6, %xmm6 ; GFNIAVX1-NEXT: vpsraw $2, %xmm6, %xmm7 -; GFNIAVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm5 -; GFNIAVX1-NEXT: vpblendvb %xmm5, %xmm7, %xmm6, %xmm6 +; GFNIAVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm8 +; GFNIAVX1-NEXT: vpblendvb %xmm8, %xmm7, %xmm6, %xmm6 ; GFNIAVX1-NEXT: vpsraw $1, %xmm6, %xmm7 -; GFNIAVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm5 +; GFNIAVX1-NEXT: vpsllw $2, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpblendvb %xmm5, %xmm7, %xmm6, %xmm5 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -2062,10 +2062,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm0, %xmm6 ; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm6, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpsraw $2, %xmm0, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm6, %xmm0, %xmm0 +; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm7 +; GFNIAVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpsraw $1, %xmm0, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; GFNIAVX1-NEXT: vpsllw $2, %xmm2, %xmm2 ; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm6, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm0, %xmm0 ; GFNIAVX1-NEXT: vpackuswb %xmm5, %xmm0, %xmm0 @@ -2078,10 +2078,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm6, %xmm7 ; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm7, %xmm6, %xmm6 ; GFNIAVX1-NEXT: vpsraw $2, %xmm6, %xmm7 -; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm4 -; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm7, %xmm6, %xmm6 +; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm8 +; GFNIAVX1-NEXT: vpblendvb %xmm8, %xmm7, %xmm6, %xmm6 ; GFNIAVX1-NEXT: vpsraw $1, %xmm6, %xmm7 -; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm4 +; GFNIAVX1-NEXT: vpsllw $2, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm7, %xmm6, %xmm4 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -2089,10 +2089,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm5, %xmm6 ; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm6, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpsraw $2, %xmm5, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm6, %xmm5, %xmm5 +; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm7 +; GFNIAVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpsraw $1, %xmm5, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; GFNIAVX1-NEXT: vpsllw $2, %xmm2, %xmm2 ; GFNIAVX1-NEXT: vpblendvb %xmm2, %xmm6, %xmm5, %xmm2 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm2, %xmm2 ; GFNIAVX1-NEXT: vpackuswb %xmm4, %xmm2, %xmm2 @@ -2102,10 +2102,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm5, %xmm6 ; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm6, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpsraw $2, %xmm5, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm4 -; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm6, %xmm5, %xmm5 +; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm7 +; GFNIAVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm5, %xmm5 ; GFNIAVX1-NEXT: vpsraw $1, %xmm5, %xmm6 -; GFNIAVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm4 +; GFNIAVX1-NEXT: vpsllw $2, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpblendvb %xmm4, %xmm6, %xmm5, %xmm4 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm4, %xmm4 ; GFNIAVX1-NEXT: vpunpcklbw {{.*#+}} xmm3 = xmm3[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -2113,10 +2113,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX1-NEXT: vpsraw $4, %xmm1, %xmm5 ; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm1, %xmm1 ; GFNIAVX1-NEXT: vpsraw $2, %xmm1, %xmm5 -; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm1, %xmm1 +; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm6 +; GFNIAVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm1, %xmm1 ; GFNIAVX1-NEXT: vpsraw $1, %xmm1, %xmm5 -; GFNIAVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 +; GFNIAVX1-NEXT: vpsllw $2, %xmm3, %xmm3 ; GFNIAVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm1, %xmm1 ; GFNIAVX1-NEXT: vpsrlw $8, %xmm1, %xmm1 ; GFNIAVX1-NEXT: vpackuswb %xmm4, %xmm1, %xmm1 @@ -2131,10 +2131,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX2-NEXT: vpsraw $4, %ymm5, %ymm6 ; GFNIAVX2-NEXT: vpblendvb %ymm4, %ymm6, %ymm5, %ymm5 ; GFNIAVX2-NEXT: vpsraw $2, %ymm5, %ymm6 -; GFNIAVX2-NEXT: vpaddw %ymm4, %ymm4, %ymm4 -; GFNIAVX2-NEXT: vpblendvb %ymm4, %ymm6, %ymm5, %ymm5 +; GFNIAVX2-NEXT: vpaddw %ymm4, %ymm4, %ymm7 +; GFNIAVX2-NEXT: vpblendvb %ymm7, %ymm6, %ymm5, %ymm5 ; GFNIAVX2-NEXT: vpsraw $1, %ymm5, %ymm6 -; GFNIAVX2-NEXT: vpaddw %ymm4, %ymm4, %ymm4 +; GFNIAVX2-NEXT: vpsllw $2, %ymm4, %ymm4 ; GFNIAVX2-NEXT: vpblendvb %ymm4, %ymm6, %ymm5, %ymm4 ; GFNIAVX2-NEXT: vpsrlw $8, %ymm4, %ymm4 ; GFNIAVX2-NEXT: vpunpcklbw {{.*#+}} ymm2 = ymm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -2142,10 +2142,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX2-NEXT: vpsraw $4, %ymm0, %ymm5 ; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm5, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpsraw $2, %ymm0, %ymm5 -; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm5, %ymm0, %ymm0 +; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm6 +; GFNIAVX2-NEXT: vpblendvb %ymm6, %ymm5, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpsraw $1, %ymm0, %ymm5 -; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; GFNIAVX2-NEXT: vpsllw $2, %ymm2, %ymm2 ; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm5, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpsrlw $8, %ymm0, %ymm0 ; GFNIAVX2-NEXT: vpackuswb %ymm4, %ymm0, %ymm0 @@ -2155,10 +2155,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX2-NEXT: vpsraw $4, %ymm4, %ymm5 ; GFNIAVX2-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm4 ; GFNIAVX2-NEXT: vpsraw $2, %ymm4, %ymm5 -; GFNIAVX2-NEXT: vpaddw %ymm3, %ymm3, %ymm3 -; GFNIAVX2-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm4 +; GFNIAVX2-NEXT: vpaddw %ymm3, %ymm3, %ymm6 +; GFNIAVX2-NEXT: vpblendvb %ymm6, %ymm5, %ymm4, %ymm4 ; GFNIAVX2-NEXT: vpsraw $1, %ymm4, %ymm5 -; GFNIAVX2-NEXT: vpaddw %ymm3, %ymm3, %ymm3 +; GFNIAVX2-NEXT: vpsllw $2, %ymm3, %ymm3 ; GFNIAVX2-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm3 ; GFNIAVX2-NEXT: vpsrlw $8, %ymm3, %ymm3 ; GFNIAVX2-NEXT: vpunpcklbw {{.*#+}} ymm2 = ymm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -2166,10 +2166,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX2-NEXT: vpsraw $4, %ymm1, %ymm4 ; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm1, %ymm1 ; GFNIAVX2-NEXT: vpsraw $2, %ymm1, %ymm4 -; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm1, %ymm1 +; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; GFNIAVX2-NEXT: vpblendvb %ymm5, %ymm4, %ymm1, %ymm1 ; GFNIAVX2-NEXT: vpsraw $1, %ymm1, %ymm4 -; GFNIAVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; GFNIAVX2-NEXT: vpsllw $2, %ymm2, %ymm2 ; GFNIAVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm1, %ymm1 ; GFNIAVX2-NEXT: vpsrlw $8, %ymm1, %ymm1 ; GFNIAVX2-NEXT: vpackuswb %ymm3, %ymm1, %ymm1 @@ -2185,10 +2185,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX512VL-NEXT: vpsraw $4, %ymm5, %ymm6 ; GFNIAVX512VL-NEXT: vpblendvb %ymm3, %ymm6, %ymm5, %ymm5 ; GFNIAVX512VL-NEXT: vpsraw $2, %ymm5, %ymm6 -; GFNIAVX512VL-NEXT: vpaddw %ymm3, %ymm3, %ymm3 -; GFNIAVX512VL-NEXT: vpblendvb %ymm3, %ymm6, %ymm5, %ymm5 +; GFNIAVX512VL-NEXT: vpaddw %ymm3, %ymm3, %ymm7 +; GFNIAVX512VL-NEXT: vpblendvb %ymm7, %ymm6, %ymm5, %ymm5 ; GFNIAVX512VL-NEXT: vpsraw $1, %ymm5, %ymm6 -; GFNIAVX512VL-NEXT: vpaddw %ymm3, %ymm3, %ymm3 +; GFNIAVX512VL-NEXT: vpsllw $2, %ymm3, %ymm3 ; GFNIAVX512VL-NEXT: vpblendvb %ymm3, %ymm6, %ymm5, %ymm3 ; GFNIAVX512VL-NEXT: vpsrlw $8, %ymm3, %ymm3 ; GFNIAVX512VL-NEXT: vpunpcklbw {{.*#+}} ymm2 = ymm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -2196,10 +2196,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX512VL-NEXT: vpsraw $4, %ymm4, %ymm5 ; GFNIAVX512VL-NEXT: vpblendvb %ymm2, %ymm5, %ymm4, %ymm4 ; GFNIAVX512VL-NEXT: vpsraw $2, %ymm4, %ymm5 -; GFNIAVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; GFNIAVX512VL-NEXT: vpblendvb %ymm2, %ymm5, %ymm4, %ymm4 +; GFNIAVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm6 +; GFNIAVX512VL-NEXT: vpblendvb %ymm6, %ymm5, %ymm4, %ymm4 ; GFNIAVX512VL-NEXT: vpsraw $1, %ymm4, %ymm5 -; GFNIAVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; GFNIAVX512VL-NEXT: vpsllw $2, %ymm2, %ymm2 ; GFNIAVX512VL-NEXT: vpblendvb %ymm2, %ymm5, %ymm4, %ymm2 ; GFNIAVX512VL-NEXT: vpsrlw $8, %ymm2, %ymm2 ; GFNIAVX512VL-NEXT: vpackuswb %ymm3, %ymm2, %ymm2 @@ -2209,10 +2209,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX512VL-NEXT: vpsraw $4, %ymm4, %ymm5 ; GFNIAVX512VL-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm4 ; GFNIAVX512VL-NEXT: vpsraw $2, %ymm4, %ymm5 -; GFNIAVX512VL-NEXT: vpaddw %ymm3, %ymm3, %ymm3 -; GFNIAVX512VL-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm4 +; GFNIAVX512VL-NEXT: vpaddw %ymm3, %ymm3, %ymm6 +; GFNIAVX512VL-NEXT: vpblendvb %ymm6, %ymm5, %ymm4, %ymm4 ; GFNIAVX512VL-NEXT: vpsraw $1, %ymm4, %ymm5 -; GFNIAVX512VL-NEXT: vpaddw %ymm3, %ymm3, %ymm3 +; GFNIAVX512VL-NEXT: vpsllw $2, %ymm3, %ymm3 ; GFNIAVX512VL-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm3 ; GFNIAVX512VL-NEXT: vpsrlw $8, %ymm3, %ymm3 ; GFNIAVX512VL-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -2220,10 +2220,10 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX512VL-NEXT: vpsraw $4, %ymm0, %ymm4 ; GFNIAVX512VL-NEXT: vpblendvb %ymm1, %ymm4, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpsraw $2, %ymm0, %ymm4 -; GFNIAVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; GFNIAVX512VL-NEXT: vpblendvb %ymm1, %ymm4, %ymm0, %ymm0 +; GFNIAVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm5 +; GFNIAVX512VL-NEXT: vpblendvb %ymm5, %ymm4, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpsraw $1, %ymm0, %ymm4 -; GFNIAVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; GFNIAVX512VL-NEXT: vpsllw $2, %ymm1, %ymm1 ; GFNIAVX512VL-NEXT: vpblendvb %ymm1, %ymm4, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpsrlw $8, %ymm0, %ymm0 ; GFNIAVX512VL-NEXT: vpackuswb %ymm3, %ymm0, %ymm0 @@ -2239,11 +2239,11 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX512BW-NEXT: vpmovb2m %zmm4, %k1 ; GFNIAVX512BW-NEXT: vmovdqu8 %zmm3, %zmm2 {%k1} ; GFNIAVX512BW-NEXT: vpsraw $2, %zmm2, %zmm3 -; GFNIAVX512BW-NEXT: vpaddw %zmm4, %zmm4, %zmm4 -; GFNIAVX512BW-NEXT: vpmovb2m %zmm4, %k1 +; GFNIAVX512BW-NEXT: vpaddw %zmm4, %zmm4, %zmm5 +; GFNIAVX512BW-NEXT: vpmovb2m %zmm5, %k1 ; GFNIAVX512BW-NEXT: vmovdqu8 %zmm3, %zmm2 {%k1} ; GFNIAVX512BW-NEXT: vpsraw $1, %zmm2, %zmm3 -; GFNIAVX512BW-NEXT: vpaddw %zmm4, %zmm4, %zmm4 +; GFNIAVX512BW-NEXT: vpsllw $2, %zmm4, %zmm4 ; GFNIAVX512BW-NEXT: vpmovb2m %zmm4, %k1 ; GFNIAVX512BW-NEXT: vmovdqu8 %zmm3, %zmm2 {%k1} ; GFNIAVX512BW-NEXT: vpsrlw $8, %zmm2, %zmm2 @@ -2253,11 +2253,11 @@ define <64 x i8> @var_ashr_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; GFNIAVX512BW-NEXT: vpmovb2m %zmm1, %k1 ; GFNIAVX512BW-NEXT: vmovdqu8 %zmm3, %zmm0 {%k1} ; GFNIAVX512BW-NEXT: vpsraw $2, %zmm0, %zmm3 -; GFNIAVX512BW-NEXT: vpaddw %zmm1, %zmm1, %zmm1 -; GFNIAVX512BW-NEXT: vpmovb2m %zmm1, %k1 +; GFNIAVX512BW-NEXT: vpaddw %zmm1, %zmm1, %zmm4 +; GFNIAVX512BW-NEXT: vpmovb2m %zmm4, %k1 ; GFNIAVX512BW-NEXT: vmovdqu8 %zmm3, %zmm0 {%k1} ; GFNIAVX512BW-NEXT: vpsraw $1, %zmm0, %zmm3 -; GFNIAVX512BW-NEXT: vpaddw %zmm1, %zmm1, %zmm1 +; GFNIAVX512BW-NEXT: vpsllw $2, %zmm1, %zmm1 ; GFNIAVX512BW-NEXT: vpmovb2m %zmm1, %k1 ; GFNIAVX512BW-NEXT: vmovdqu8 %zmm3, %zmm0 {%k1} ; GFNIAVX512BW-NEXT: vpsrlw $8, %zmm0, %zmm0 diff --git a/llvm/test/CodeGen/X86/logic-shift.ll b/llvm/test/CodeGen/X86/logic-shift.ll index 96e63d1..104151c 100644 --- a/llvm/test/CodeGen/X86/logic-shift.ll +++ b/llvm/test/CodeGen/X86/logic-shift.ll @@ -129,10 +129,10 @@ define <16 x i8> @or_ashr_commute3(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %y, < ; CHECK-NEXT: vpsraw $4, %xmm1, %xmm5 ; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsraw $2, %xmm1, %xmm5 -; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm4 -; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 +; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm6 +; CHECK-NEXT: vpblendvb %xmm6, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsraw $1, %xmm1, %xmm5 -; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm4 +; CHECK-NEXT: vpsllw $2, %xmm4, %xmm4 ; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsrlw $8, %xmm1, %xmm1 ; CHECK-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -140,10 +140,10 @@ define <16 x i8> @or_ashr_commute3(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %y, < ; CHECK-NEXT: vpsraw $4, %xmm0, %xmm4 ; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsraw $2, %xmm0, %xmm4 -; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 +; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; CHECK-NEXT: vpblendvb %xmm5, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsraw $1, %xmm0, %xmm4 -; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; CHECK-NEXT: vpsllw $2, %xmm2, %xmm2 ; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsrlw $8, %xmm0, %xmm0 ; CHECK-NEXT: vpackuswb %xmm1, %xmm0, %xmm0 @@ -413,10 +413,10 @@ define <16 x i8> @xor_ashr_commute3(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %y, ; CHECK-NEXT: vpsraw $4, %xmm1, %xmm5 ; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsraw $2, %xmm1, %xmm5 -; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm4 -; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 +; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm6 +; CHECK-NEXT: vpblendvb %xmm6, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsraw $1, %xmm1, %xmm5 -; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm4 +; CHECK-NEXT: vpsllw $2, %xmm4, %xmm4 ; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsrlw $8, %xmm1, %xmm1 ; CHECK-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -424,10 +424,10 @@ define <16 x i8> @xor_ashr_commute3(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %y, ; CHECK-NEXT: vpsraw $4, %xmm0, %xmm4 ; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsraw $2, %xmm0, %xmm4 -; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 +; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; CHECK-NEXT: vpblendvb %xmm5, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsraw $1, %xmm0, %xmm4 -; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; CHECK-NEXT: vpsllw $2, %xmm2, %xmm2 ; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsrlw $8, %xmm0, %xmm0 ; CHECK-NEXT: vpackuswb %xmm1, %xmm0, %xmm0 @@ -697,10 +697,10 @@ define <16 x i8> @and_ashr_commute3(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %y, ; CHECK-NEXT: vpsraw $4, %xmm1, %xmm5 ; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsraw $2, %xmm1, %xmm5 -; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm4 -; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 +; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm6 +; CHECK-NEXT: vpblendvb %xmm6, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsraw $1, %xmm1, %xmm5 -; CHECK-NEXT: vpaddw %xmm4, %xmm4, %xmm4 +; CHECK-NEXT: vpsllw $2, %xmm4, %xmm4 ; CHECK-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 ; CHECK-NEXT: vpsrlw $8, %xmm1, %xmm1 ; CHECK-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -708,10 +708,10 @@ define <16 x i8> @and_ashr_commute3(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %y, ; CHECK-NEXT: vpsraw $4, %xmm0, %xmm4 ; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsraw $2, %xmm0, %xmm4 -; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 +; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; CHECK-NEXT: vpblendvb %xmm5, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsraw $1, %xmm0, %xmm4 -; CHECK-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; CHECK-NEXT: vpsllw $2, %xmm2, %xmm2 ; CHECK-NEXT: vpblendvb %xmm2, %xmm4, %xmm0, %xmm0 ; CHECK-NEXT: vpsrlw $8, %xmm0, %xmm0 ; CHECK-NEXT: vpackuswb %xmm1, %xmm0, %xmm0 diff --git a/llvm/test/CodeGen/X86/prefer-avx256-shift.ll b/llvm/test/CodeGen/X86/prefer-avx256-shift.ll index bf04c8d..63bbac12 100644 --- a/llvm/test/CodeGen/X86/prefer-avx256-shift.ll +++ b/llvm/test/CodeGen/X86/prefer-avx256-shift.ll @@ -302,10 +302,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) { ; AVX256-NEXT: vpsraw $4, %ymm3, %ymm4 ; AVX256-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; AVX256-NEXT: vpsraw $2, %ymm3, %ymm4 -; AVX256-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; AVX256-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; AVX256-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; AVX256-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; AVX256-NEXT: vpsraw $1, %ymm3, %ymm4 -; AVX256-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; AVX256-NEXT: vpsllw $2, %ymm2, %ymm2 ; AVX256-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; AVX256-NEXT: vpsrlw $8, %ymm2, %ymm2 ; AVX256-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -313,10 +313,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) { ; AVX256-NEXT: vpsraw $4, %ymm0, %ymm3 ; AVX256-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX256-NEXT: vpsraw $2, %ymm0, %ymm3 -; AVX256-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; AVX256-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; AVX256-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; AVX256-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; AVX256-NEXT: vpsraw $1, %ymm0, %ymm3 -; AVX256-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; AVX256-NEXT: vpsllw $2, %ymm1, %ymm1 ; AVX256-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX256-NEXT: vpsrlw $8, %ymm0, %ymm0 ; AVX256-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 @@ -338,10 +338,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) { ; AVX512VL-NEXT: vpsraw $4, %ymm3, %ymm4 ; AVX512VL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; AVX512VL-NEXT: vpsraw $2, %ymm3, %ymm4 -; AVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; AVX512VL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; AVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; AVX512VL-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; AVX512VL-NEXT: vpsraw $1, %ymm3, %ymm4 -; AVX512VL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; AVX512VL-NEXT: vpsllw $2, %ymm2, %ymm2 ; AVX512VL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; AVX512VL-NEXT: vpsrlw $8, %ymm2, %ymm2 ; AVX512VL-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -349,10 +349,10 @@ define <32 x i8> @var_ashr_v32i8(<32 x i8> %a, <32 x i8> %b) { ; AVX512VL-NEXT: vpsraw $4, %ymm0, %ymm3 ; AVX512VL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX512VL-NEXT: vpsraw $2, %ymm0, %ymm3 -; AVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; AVX512VL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; AVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; AVX512VL-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; AVX512VL-NEXT: vpsraw $1, %ymm0, %ymm3 -; AVX512VL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; AVX512VL-NEXT: vpsllw $2, %ymm1, %ymm1 ; AVX512VL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX512VL-NEXT: vpsrlw $8, %ymm0, %ymm0 ; AVX512VL-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 @@ -432,10 +432,10 @@ define <16 x i8> @var_ashr_v16i8(<16 x i8> %a, <16 x i8> %b) { ; AVX256VL-NEXT: vpsraw $4, %xmm3, %xmm4 ; AVX256VL-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 ; AVX256VL-NEXT: vpsraw $2, %xmm3, %xmm4 -; AVX256VL-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX256VL-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 +; AVX256VL-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; AVX256VL-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; AVX256VL-NEXT: vpsraw $1, %xmm3, %xmm4 -; AVX256VL-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; AVX256VL-NEXT: vpsllw $2, %xmm2, %xmm2 ; AVX256VL-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; AVX256VL-NEXT: vpsrlw $8, %xmm2, %xmm2 ; AVX256VL-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -443,10 +443,10 @@ define <16 x i8> @var_ashr_v16i8(<16 x i8> %a, <16 x i8> %b) { ; AVX256VL-NEXT: vpsraw $4, %xmm0, %xmm3 ; AVX256VL-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX256VL-NEXT: vpsraw $2, %xmm0, %xmm3 -; AVX256VL-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; AVX256VL-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 +; AVX256VL-NEXT: vpaddw %xmm1, %xmm1, %xmm4 +; AVX256VL-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 ; AVX256VL-NEXT: vpsraw $1, %xmm0, %xmm3 -; AVX256VL-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; AVX256VL-NEXT: vpsllw $2, %xmm1, %xmm1 ; AVX256VL-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX256VL-NEXT: vpsrlw $8, %xmm0, %xmm0 ; AVX256VL-NEXT: vpackuswb %xmm2, %xmm0, %xmm0 diff --git a/llvm/test/CodeGen/X86/shuffle-as-shifts.ll b/llvm/test/CodeGen/X86/shuffle-as-shifts.ll index 9c8729b3..4b8f78d 100644 --- a/llvm/test/CodeGen/X86/shuffle-as-shifts.ll +++ b/llvm/test/CodeGen/X86/shuffle-as-shifts.ll @@ -15,20 +15,20 @@ define <4 x i32> @shuf_rot_v4i32_1032(<4 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_rot_v4i32_1032: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2] +; CHECK-ICX-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_rot_v4i32_1032: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2] +; CHECK-V4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_rot_v4i32_1032: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2] +; CHECK-ZNVER4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <4 x i32> %x, %x %r = shufflevector <4 x i32> %x1, <4 x i32> zeroinitializer, <4 x i32> <i32 1, i32 0, i32 3, i32 2> @@ -44,20 +44,20 @@ define <8 x i32> @shuf_rot_v8i32_10325476(<8 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_rot_v8i32_10325476: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6] +; CHECK-ICX-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_rot_v8i32_10325476: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6] +; CHECK-V4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_rot_v8i32_10325476: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6] +; CHECK-ZNVER4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <8 x i32> %x, %x %r = shufflevector <8 x i32> %x1, <8 x i32> zeroinitializer, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6> @@ -73,20 +73,20 @@ define <16 x i32> @shuf_rot_v16i32_1032547698111013121514(<16 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_rot_v16i32_1032547698111013121514: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14] +; CHECK-ICX-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_rot_v16i32_1032547698111013121514: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14] +; CHECK-V4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_rot_v16i32_1032547698111013121514: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14] +; CHECK-ZNVER4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <16 x i32> %x, %x %r = shufflevector <16 x i32> %x1, <16 x i32> zeroinitializer, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14> @@ -168,20 +168,20 @@ define <4 x i32> @shuf_shr_v4i32_1U3U(<4 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_shr_v4i32_1U3U: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; CHECK-ICX-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_shr_v4i32_1U3U: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; CHECK-V4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_shr_v4i32_1U3U: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; CHECK-ZNVER4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <4 x i32> %x, %x %r = shufflevector <4 x i32> %x1, <4 x i32> zeroinitializer, <4 x i32> <i32 1, i32 undef, i32 3, i32 undef> @@ -197,20 +197,20 @@ define <8 x i32> @shuf_shr_v8i32_1U3U5U7U(<8 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_shr_v8i32_1U3U5U7U: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,1,3,3,5,5,7,7] +; CHECK-ICX-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_shr_v8i32_1U3U5U7U: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,1,3,3,5,5,7,7] +; CHECK-V4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_shr_v8i32_1U3U5U7U: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,1,3,3,5,5,7,7] +; CHECK-ZNVER4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <8 x i32> %x, %x %r = shufflevector <8 x i32> %x1, <8 x i32> zeroinitializer, <8 x i32> <i32 1, i32 undef, i32 3, i32 undef, i32 5, i32 undef, i32 7, i32 undef> @@ -226,20 +226,20 @@ define <16 x i32> @shuf_shr_v16i32_U3U5U7U9U11U13U15(<16 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_shr_v16i32_U3U5U7U9U11U13U15: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15] +; CHECK-ICX-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_shr_v16i32_U3U5U7U9U11U13U15: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15] +; CHECK-V4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_shr_v16i32_U3U5U7U9U11U13U15: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15] +; CHECK-ZNVER4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <16 x i32> %x, %x %r = shufflevector <16 x i32> %x1, <16 x i32> zeroinitializer, <16 x i32> <i32 1, i32 undef, i32 3, i32 undef, i32 5, i32 undef, i32 7, i32 undef, i32 9, i32 undef, i32 11, i32 undef, i32 13, i32 undef, i32 15, i32 undef> @@ -288,20 +288,20 @@ define <4 x i32> @shuf_shl_v4i32_U0U2(<4 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_shl_v4i32_U0U2: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,2,2] +; CHECK-ICX-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_shl_v4i32_U0U2: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,2,2] +; CHECK-V4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_shl_v4i32_U0U2: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,2,2] +; CHECK-ZNVER4-NEXT: vpaddd %xmm0, %xmm0, %xmm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <4 x i32> %x, %x %r = shufflevector <4 x i32> %x1, <4 x i32> zeroinitializer, <4 x i32> <i32 undef, i32 0, i32 undef, i32 2> @@ -317,20 +317,20 @@ define <8 x i32> @shuf_shl_v8i32_U0U2U4U6(<8 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_shl_v8i32_U0U2U4U6: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,0,2,2,4,4,6,6] +; CHECK-ICX-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_shl_v8i32_U0U2U4U6: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,0,2,2,4,4,6,6] +; CHECK-V4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_shl_v8i32_U0U2U4U6: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,0,2,2,4,4,6,6] +; CHECK-ZNVER4-NEXT: vpaddd %ymm0, %ymm0, %ymm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <8 x i32> %x, %x %r = shufflevector <8 x i32> %x1, <8 x i32> zeroinitializer, <8 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 undef, i32 4, i32 undef, i32 6> @@ -346,20 +346,20 @@ define <16 x i32> @shuf_shl_v16i32_U0U2U4U6U8U10U12U14(<16 x i32> %x) { ; ; CHECK-ICX-LABEL: shuf_shl_v16i32_U0U2U4U6U8U10U12U14: ; CHECK-ICX: # %bb.0: -; CHECK-ICX-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ICX-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14] +; CHECK-ICX-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ICX-NEXT: retq ; ; CHECK-V4-LABEL: shuf_shl_v16i32_U0U2U4U6U8U10U12U14: ; CHECK-V4: # %bb.0: -; CHECK-V4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-V4-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14] +; CHECK-V4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-V4-NEXT: retq ; ; CHECK-ZNVER4-LABEL: shuf_shl_v16i32_U0U2U4U6U8U10U12U14: ; CHECK-ZNVER4: # %bb.0: -; CHECK-ZNVER4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ZNVER4-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14] +; CHECK-ZNVER4-NEXT: vpaddd %zmm0, %zmm0, %zmm0 ; CHECK-ZNVER4-NEXT: retq %x1 = add <16 x i32> %x, %x %r = shufflevector <16 x i32> %x1, <16 x i32> zeroinitializer, <16 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 undef, i32 4, i32 undef, i32 6, i32 undef, i32 8, i32 undef, i32 10, i32 undef, i32 12, i32 undef, i32 14> diff --git a/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll b/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll index ecd9435..1766b4d 100644 --- a/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll +++ b/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll @@ -58,7 +58,7 @@ define <8 x float> @foo8(<8 x float> %v, ptr%p) nounwind { define <4 x i32> @undef_splatmask(<4 x i32> %v) nounwind { ; AVX2-LABEL: undef_splatmask: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,2,2] ; AVX2-NEXT: retq %res = shufflevector <4 x i32> %v, <4 x i32> undef, <4 x i32> <i32 2, i32 undef, i32 2, i32 undef> %res1 = shufflevector <4 x i32> %res, <4 x i32> undef, <4 x i32> <i32 0, i32 2, i32 undef, i32 undef> @@ -68,7 +68,7 @@ define <4 x i32> @undef_splatmask(<4 x i32> %v) nounwind { define <4 x i32> @undef_splatmask2(<4 x i32> %v) nounwind { ; AVX2-LABEL: undef_splatmask2: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,2,2] ; AVX2-NEXT: retq %res = shufflevector <4 x i32> %v, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 2, i32 undef> %res1 = shufflevector <4 x i32> %res, <4 x i32> undef, <4 x i32> <i32 0, i32 2, i32 undef, i32 undef> @@ -78,7 +78,7 @@ define <4 x i32> @undef_splatmask2(<4 x i32> %v) nounwind { define <4 x i32> @undef_splatmask3(<4 x i32> %v) nounwind { ; AVX2-LABEL: undef_splatmask3: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,2,2] ; AVX2-NEXT: retq %res = shufflevector <4 x i32> %v, <4 x i32> undef, <4 x i32> <i32 2, i32 undef, i32 2, i32 undef> %res1 = shufflevector <4 x i32> %res, <4 x i32> undef, <4 x i32> <i32 0, i32 2, i32 undef, i32 3> @@ -88,7 +88,7 @@ define <4 x i32> @undef_splatmask3(<4 x i32> %v) nounwind { define <4 x i32> @undef_splatmask4(<4 x i32> %v, ptr %p) nounwind { ; AVX2-LABEL: undef_splatmask4: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,2,2,2] ; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,3,2,3] ; AVX2-NEXT: vmovaps %xmm0, (%rdi) ; AVX2-NEXT: vmovaps %xmm1, %xmm0 diff --git a/llvm/test/CodeGen/X86/sshl_sat_vec.ll b/llvm/test/CodeGen/X86/sshl_sat_vec.ll index f91758b..10dee14 100644 --- a/llvm/test/CodeGen/X86/sshl_sat_vec.ll +++ b/llvm/test/CodeGen/X86/sshl_sat_vec.ll @@ -602,10 +602,10 @@ define <16 x i8> @vec_v16i8(<16 x i8> %x, <16 x i8> %y) nounwind { ; X64-AVX2-NEXT: vpunpckhbw {{.*#+}} xmm5 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15] ; X64-AVX2-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; X64-AVX2-NEXT: vpsraw $2, %xmm3, %xmm4 -; X64-AVX2-NEXT: vpaddw %xmm5, %xmm5, %xmm5 -; X64-AVX2-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 +; X64-AVX2-NEXT: vpaddw %xmm5, %xmm5, %xmm6 +; X64-AVX2-NEXT: vpblendvb %xmm6, %xmm4, %xmm3, %xmm3 ; X64-AVX2-NEXT: vpsraw $1, %xmm3, %xmm4 -; X64-AVX2-NEXT: vpaddw %xmm5, %xmm5, %xmm5 +; X64-AVX2-NEXT: vpsllw $2, %xmm5, %xmm5 ; X64-AVX2-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; X64-AVX2-NEXT: vpsrlw $8, %xmm3, %xmm3 ; X64-AVX2-NEXT: vpunpcklbw {{.*#+}} xmm4 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -613,10 +613,10 @@ define <16 x i8> @vec_v16i8(<16 x i8> %x, <16 x i8> %y) nounwind { ; X64-AVX2-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] ; X64-AVX2-NEXT: vpblendvb %xmm1, %xmm5, %xmm4, %xmm4 ; X64-AVX2-NEXT: vpsraw $2, %xmm4, %xmm5 -; X64-AVX2-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; X64-AVX2-NEXT: vpblendvb %xmm1, %xmm5, %xmm4, %xmm4 +; X64-AVX2-NEXT: vpaddw %xmm1, %xmm1, %xmm6 +; X64-AVX2-NEXT: vpblendvb %xmm6, %xmm5, %xmm4, %xmm4 ; X64-AVX2-NEXT: vpsraw $1, %xmm4, %xmm5 -; X64-AVX2-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; X64-AVX2-NEXT: vpsllw $2, %xmm1, %xmm1 ; X64-AVX2-NEXT: vpblendvb %xmm1, %xmm5, %xmm4, %xmm1 ; X64-AVX2-NEXT: vpsrlw $8, %xmm1, %xmm1 ; X64-AVX2-NEXT: vpackuswb %xmm3, %xmm1, %xmm1 diff --git a/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll b/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll index 209d6a5..93a692c 100644 --- a/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll +++ b/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll @@ -1911,13 +1911,13 @@ define <2 x i64> @test_v2f64_ogt_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmoval %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmoval %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2031,13 +2031,13 @@ define <2 x i64> @test_v2f64_oge_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovael %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmovael %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2151,13 +2151,13 @@ define <2 x i64> @test_v2f64_olt_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmoval %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmoval %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2269,13 +2269,13 @@ define <2 x i64> @test_v2f64_ole_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovael %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmovael %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2680,13 +2680,13 @@ define <2 x i64> @test_v2f64_ugt_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbl %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmovbl %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2798,13 +2798,13 @@ define <2 x i64> @test_v2f64_uge_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbel %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmovbel %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2916,13 +2916,13 @@ define <2 x i64> @test_v2f64_ult_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbl %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmovbl %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -3036,13 +3036,13 @@ define <2 x i64> @test_v2f64_ule_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbel %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmovbel %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 diff --git a/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll b/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll index 9ecc629..b378dce 100644 --- a/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll @@ -162,7 +162,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE2-NEXT: pslld $23, %xmm1 ; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 ; SSE2-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 @@ -182,7 +182,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3] ; SSE41-NEXT: pslld $23, %xmm1 ; SSE41-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 @@ -200,7 +200,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[1,1,3,3] ; AVX1-NEXT: vpslld $23, %xmm1, %xmm1 ; AVX1-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 @@ -277,7 +277,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; XOPAVX1-NEXT: vprotd %xmm1, %xmm0, %xmm0 ; XOPAVX1-NEXT: retq ; @@ -289,7 +289,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; X86-SSE2-NEXT: pslld $23, %xmm1 ; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1 ; X86-SSE2-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1 diff --git a/llvm/test/CodeGen/X86/vector-fshl-sub128.ll b/llvm/test/CodeGen/X86/vector-fshl-sub128.ll index 322ebe2..06ff7e7 100644 --- a/llvm/test/CodeGen/X86/vector-fshl-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshl-sub128.ll @@ -250,7 +250,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; SSE2-NEXT: movdqa %xmm3, %xmm5 ; SSE2-NEXT: pandn %xmm4, %xmm5 @@ -286,7 +286,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE41-NEXT: pmovsxbd {{.*#+}} xmm3 = [31,31,31,31] ; SSE41-NEXT: movdqa %xmm2, %xmm4 ; SSE41-NEXT: pandn %xmm3, %xmm4 @@ -316,7 +316,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; AVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; AVX1-NEXT: vpandn %xmm3, %xmm2, %xmm4 ; AVX1-NEXT: vpsrldq {{.*#+}} xmm5 = xmm4[12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero @@ -423,7 +423,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; XOPAVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; XOPAVX1-NEXT: vpand %xmm3, %xmm2, %xmm4 ; XOPAVX1-NEXT: vpshld %xmm4, %xmm0, %xmm0 @@ -450,7 +450,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; X86-SSE2-NEXT: movdqa %xmm3, %xmm5 ; X86-SSE2-NEXT: pandn %xmm4, %xmm5 diff --git a/llvm/test/CodeGen/X86/vector-fshr-128.ll b/llvm/test/CodeGen/X86/vector-fshr-128.ll index 20be579..9b52857 100644 --- a/llvm/test/CodeGen/X86/vector-fshr-128.ll +++ b/llvm/test/CodeGen/X86/vector-fshr-128.ll @@ -536,14 +536,14 @@ define <8 x i16> @var_funnnel_v8i16(<8 x i16> %x, <8 x i16> %y, <8 x i16> %amt) ; AVX1-NEXT: vpaddw %xmm4, %xmm4, %xmm5 ; AVX1-NEXT: vpsrlw $8, %xmm1, %xmm6 ; AVX1-NEXT: vpblendvb %xmm4, %xmm6, %xmm1, %xmm1 -; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm4 -; AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm1, %xmm1 -; AVX1-NEXT: vpsrlw $2, %xmm1, %xmm4 -; AVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm5 -; AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm1, %xmm1 -; AVX1-NEXT: vpsrlw $1, %xmm1, %xmm4 -; AVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm5 -; AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm1, %xmm1 +; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm6 +; AVX1-NEXT: vpblendvb %xmm5, %xmm6, %xmm1, %xmm1 +; AVX1-NEXT: vpsrlw $2, %xmm1, %xmm5 +; AVX1-NEXT: vpsllw $2, %xmm4, %xmm6 +; AVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm1, %xmm1 +; AVX1-NEXT: vpsrlw $1, %xmm1, %xmm5 +; AVX1-NEXT: vpsllw $3, %xmm4, %xmm4 +; AVX1-NEXT: vpblendvb %xmm4, %xmm5, %xmm1, %xmm1 ; AVX1-NEXT: vpandn %xmm3, %xmm2, %xmm2 ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm3 = xmm2[4,4,5,5,6,6,7,7] ; AVX1-NEXT: vpslld $23, %xmm3, %xmm3 diff --git a/llvm/test/CodeGen/X86/vector-fshr-256.ll b/llvm/test/CodeGen/X86/vector-fshr-256.ll index 1f16463..a387562 100644 --- a/llvm/test/CodeGen/X86/vector-fshr-256.ll +++ b/llvm/test/CodeGen/X86/vector-fshr-256.ll @@ -328,15 +328,15 @@ define <16 x i16> @var_funnnel_v16i16(<16 x i16> %x, <16 x i16> %y, <16 x i16> % ; AVX1-NEXT: vpaddw %xmm5, %xmm5, %xmm6 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm7 ; AVX1-NEXT: vpsrlw $8, %xmm7, %xmm8 -; AVX1-NEXT: vpblendvb %xmm5, %xmm8, %xmm7, %xmm5 -; AVX1-NEXT: vpsrlw $4, %xmm5, %xmm7 -; AVX1-NEXT: vpblendvb %xmm6, %xmm7, %xmm5, %xmm5 -; AVX1-NEXT: vpsrlw $2, %xmm5, %xmm7 -; AVX1-NEXT: vpaddw %xmm6, %xmm6, %xmm6 -; AVX1-NEXT: vpblendvb %xmm6, %xmm7, %xmm5, %xmm5 -; AVX1-NEXT: vpsrlw $1, %xmm5, %xmm7 -; AVX1-NEXT: vpaddw %xmm6, %xmm6, %xmm6 -; AVX1-NEXT: vpblendvb %xmm6, %xmm7, %xmm5, %xmm5 +; AVX1-NEXT: vpblendvb %xmm5, %xmm8, %xmm7, %xmm7 +; AVX1-NEXT: vpsrlw $4, %xmm7, %xmm8 +; AVX1-NEXT: vpblendvb %xmm6, %xmm8, %xmm7, %xmm6 +; AVX1-NEXT: vpsrlw $2, %xmm6, %xmm7 +; AVX1-NEXT: vpsllw $2, %xmm5, %xmm8 +; AVX1-NEXT: vpblendvb %xmm8, %xmm7, %xmm6, %xmm6 +; AVX1-NEXT: vpsrlw $1, %xmm6, %xmm7 +; AVX1-NEXT: vpsllw $3, %xmm5, %xmm5 +; AVX1-NEXT: vpblendvb %xmm5, %xmm7, %xmm6, %xmm5 ; AVX1-NEXT: vpxor %xmm3, %xmm4, %xmm6 ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm4 = xmm6[4,4,5,5,6,6,7,7] ; AVX1-NEXT: vpslld $23, %xmm4, %xmm7 @@ -358,14 +358,14 @@ define <16 x i16> @var_funnnel_v16i16(<16 x i16> %x, <16 x i16> %y, <16 x i16> % ; AVX1-NEXT: vpaddw %xmm6, %xmm6, %xmm7 ; AVX1-NEXT: vpsrlw $8, %xmm1, %xmm8 ; AVX1-NEXT: vpblendvb %xmm6, %xmm8, %xmm1, %xmm1 -; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm6 -; AVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm1, %xmm1 -; AVX1-NEXT: vpsrlw $2, %xmm1, %xmm6 -; AVX1-NEXT: vpaddw %xmm7, %xmm7, %xmm7 -; AVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm1, %xmm1 -; AVX1-NEXT: vpsrlw $1, %xmm1, %xmm6 -; AVX1-NEXT: vpaddw %xmm7, %xmm7, %xmm7 -; AVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm1, %xmm1 +; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm8 +; AVX1-NEXT: vpblendvb %xmm7, %xmm8, %xmm1, %xmm1 +; AVX1-NEXT: vpsrlw $2, %xmm1, %xmm7 +; AVX1-NEXT: vpsllw $2, %xmm6, %xmm8 +; AVX1-NEXT: vpblendvb %xmm8, %xmm7, %xmm1, %xmm1 +; AVX1-NEXT: vpsrlw $1, %xmm1, %xmm7 +; AVX1-NEXT: vpsllw $3, %xmm6, %xmm6 +; AVX1-NEXT: vpblendvb %xmm6, %xmm7, %xmm1, %xmm1 ; AVX1-NEXT: vpxor %xmm3, %xmm2, %xmm2 ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm3 = xmm2[4,4,5,5,6,6,7,7] ; AVX1-NEXT: vpslld $23, %xmm3, %xmm3 diff --git a/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll b/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll index 178c02f..ef5ffe4 100644 --- a/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll @@ -172,7 +172,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE2-NEXT: pxor %xmm2, %xmm2 ; SSE2-NEXT: psubd %xmm1, %xmm2 ; SSE2-NEXT: pslld $23, %xmm2 @@ -194,7 +194,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE41-NEXT: pxor %xmm2, %xmm2 ; SSE41-NEXT: psubd %xmm1, %xmm2 ; SSE41-NEXT: pslld $23, %xmm2 @@ -214,7 +214,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2 ; AVX1-NEXT: vpsubd %xmm1, %xmm2, %xmm1 ; AVX1-NEXT: vpslld $23, %xmm1, %xmm1 @@ -293,7 +293,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; XOPAVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2 ; XOPAVX1-NEXT: vpsubd %xmm1, %xmm2, %xmm1 ; XOPAVX1-NEXT: vprotd %xmm1, %xmm0, %xmm0 @@ -309,7 +309,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; X86-SSE2-NEXT: pxor %xmm2, %xmm2 ; X86-SSE2-NEXT: psubd %xmm1, %xmm2 ; X86-SSE2-NEXT: pslld $23, %xmm2 diff --git a/llvm/test/CodeGen/X86/vector-fshr-sub128.ll b/llvm/test/CodeGen/X86/vector-fshr-sub128.ll index 372deb05..2d8670a 100644 --- a/llvm/test/CodeGen/X86/vector-fshr-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshr-sub128.ll @@ -251,7 +251,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; SSE2-NEXT: movdqa %xmm3, %xmm5 ; SSE2-NEXT: pand %xmm4, %xmm5 @@ -287,7 +287,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE41-NEXT: pmovsxbd {{.*#+}} xmm3 = [31,31,31,31] ; SSE41-NEXT: movdqa %xmm2, %xmm4 ; SSE41-NEXT: pand %xmm3, %xmm4 @@ -317,7 +317,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; AVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm4 ; AVX1-NEXT: vpsrldq {{.*#+}} xmm5 = xmm4[12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero @@ -425,7 +425,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; XOPAVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; XOPAVX1-NEXT: vpandn %xmm3, %xmm2, %xmm4 ; XOPAVX1-NEXT: vpaddd %xmm0, %xmm0, %xmm0 @@ -452,7 +452,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; X86-SSE2-NEXT: movdqa %xmm3, %xmm5 ; X86-SSE2-NEXT: pand %xmm4, %xmm5 diff --git a/llvm/test/CodeGen/X86/vector-sext.ll b/llvm/test/CodeGen/X86/vector-sext.ll index f57efb4..1e11ea9 100644 --- a/llvm/test/CodeGen/X86/vector-sext.ll +++ b/llvm/test/CodeGen/X86/vector-sext.ll @@ -1409,11 +1409,11 @@ define <2 x i64> @load_sext_2i1_to_2i64(ptr%ptr) { ; X86-SSE2-NEXT: movzbl %al, %eax ; X86-SSE2-NEXT: negl %eax ; X86-SSE2-NEXT: movd %eax, %xmm0 -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,0,0] ; X86-SSE2-NEXT: andl $1, %ecx ; X86-SSE2-NEXT: negl %ecx ; X86-SSE2-NEXT: movd %ecx, %xmm0 -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; X86-SSE2-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] ; X86-SSE2-NEXT: retl ; diff --git a/llvm/test/CodeGen/X86/vector-shift-ashr-128.ll b/llvm/test/CodeGen/X86/vector-shift-ashr-128.ll index 02f0f53..d565ef0 100644 --- a/llvm/test/CodeGen/X86/vector-shift-ashr-128.ll +++ b/llvm/test/CodeGen/X86/vector-shift-ashr-128.ll @@ -293,14 +293,14 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm2 ; AVX1-NEXT: vpsraw $8, %xmm0, %xmm3 ; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $4, %xmm0, %xmm3 +; AVX1-NEXT: vpblendvb %xmm2, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $2, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm3 +; AVX1-NEXT: vpblendvb %xmm3, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $1, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: retq ; ; AVX2-LABEL: var_shift_v8i16: @@ -494,10 +494,10 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm3, %xmm4 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $2, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 +; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; AVX-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $1, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; AVX-NEXT: vpsllw $2, %xmm2, %xmm2 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; AVX-NEXT: vpsrlw $8, %xmm2, %xmm2 ; AVX-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -505,10 +505,10 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm0, %xmm3 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $2, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 +; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm4 +; AVX-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $1, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpsllw $2, %xmm1, %xmm1 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0 ; AVX-NEXT: vpackuswb %xmm2, %xmm0, %xmm0 diff --git a/llvm/test/CodeGen/X86/vector-shift-ashr-256.ll b/llvm/test/CodeGen/X86/vector-shift-ashr-256.ll index 15855e3..249bcba 100644 --- a/llvm/test/CodeGen/X86/vector-shift-ashr-256.ll +++ b/llvm/test/CodeGen/X86/vector-shift-ashr-256.ll @@ -237,29 +237,29 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm3 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm4 ; AVX1-NEXT: vpsraw $8, %xmm4, %xmm5 -; AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm2 -; AVX1-NEXT: vpsraw $4, %xmm2, %xmm4 -; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; AVX1-NEXT: vpsraw $2, %xmm2, %xmm4 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; AVX1-NEXT: vpsraw $1, %xmm2, %xmm4 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 +; AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 +; AVX1-NEXT: vpsraw $4, %xmm4, %xmm5 +; AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm3 +; AVX1-NEXT: vpsraw $2, %xmm3, %xmm4 +; AVX1-NEXT: vpsllw $2, %xmm2, %xmm5 +; AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vpsraw $1, %xmm3, %xmm4 +; AVX1-NEXT: vpsllw $3, %xmm2, %xmm2 +; AVX1-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; AVX1-NEXT: vpsllw $12, %xmm1, %xmm3 ; AVX1-NEXT: vpsllw $4, %xmm1, %xmm1 ; AVX1-NEXT: vpor %xmm3, %xmm1, %xmm1 ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm3 ; AVX1-NEXT: vpsraw $8, %xmm0, %xmm4 ; AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $4, %xmm0, %xmm4 +; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $2, %xmm0, %xmm3 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm4 +; AVX1-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $1, %xmm0, %xmm3 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: retq ; @@ -339,29 +339,29 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) nounwind { ; X86-AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm3 ; X86-AVX1-NEXT: vextractf128 $1, %ymm0, %xmm4 ; X86-AVX1-NEXT: vpsraw $8, %xmm4, %xmm5 -; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm2 -; X86-AVX1-NEXT: vpsraw $4, %xmm2, %xmm4 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; X86-AVX1-NEXT: vpsraw $2, %xmm2, %xmm4 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; X86-AVX1-NEXT: vpsraw $1, %xmm2, %xmm4 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 +; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 +; X86-AVX1-NEXT: vpsraw $4, %xmm4, %xmm5 +; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm3 +; X86-AVX1-NEXT: vpsraw $2, %xmm3, %xmm4 +; X86-AVX1-NEXT: vpsllw $2, %xmm2, %xmm5 +; X86-AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 +; X86-AVX1-NEXT: vpsraw $1, %xmm3, %xmm4 +; X86-AVX1-NEXT: vpsllw $3, %xmm2, %xmm2 +; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; X86-AVX1-NEXT: vpsllw $12, %xmm1, %xmm3 ; X86-AVX1-NEXT: vpsllw $4, %xmm1, %xmm1 ; X86-AVX1-NEXT: vpor %xmm3, %xmm1, %xmm1 ; X86-AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm3 ; X86-AVX1-NEXT: vpsraw $8, %xmm0, %xmm4 ; X86-AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 -; X86-AVX1-NEXT: vpsraw $4, %xmm0, %xmm1 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; X86-AVX1-NEXT: vpsraw $2, %xmm0, %xmm1 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; X86-AVX1-NEXT: vpsraw $1, %xmm0, %xmm1 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 +; X86-AVX1-NEXT: vpsraw $4, %xmm0, %xmm4 +; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm0, %xmm0 +; X86-AVX1-NEXT: vpsraw $2, %xmm0, %xmm3 +; X86-AVX1-NEXT: vpsllw $2, %xmm1, %xmm4 +; X86-AVX1-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 +; X86-AVX1-NEXT: vpsraw $1, %xmm0, %xmm3 +; X86-AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; X86-AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; X86-AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; X86-AVX1-NEXT: retl ; @@ -393,10 +393,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX1-NEXT: vpsraw $4, %xmm5, %xmm6 ; AVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm5 ; AVX1-NEXT: vpsraw $2, %xmm5, %xmm6 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm5 +; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm7 +; AVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm5, %xmm5 ; AVX1-NEXT: vpsraw $1, %xmm5, %xmm6 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 +; AVX1-NEXT: vpsllw $2, %xmm3, %xmm3 ; AVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm3 ; AVX1-NEXT: vpsrlw $8, %xmm3, %xmm3 ; AVX1-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -404,10 +404,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX1-NEXT: vpsraw $4, %xmm4, %xmm5 ; AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 ; AVX1-NEXT: vpsraw $2, %xmm4, %xmm5 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 +; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm6 +; AVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm4, %xmm4 ; AVX1-NEXT: vpsraw $1, %xmm4, %xmm5 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm2, %xmm2 ; AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm2 ; AVX1-NEXT: vpsrlw $8, %xmm2, %xmm2 ; AVX1-NEXT: vpackuswb %xmm3, %xmm2, %xmm2 @@ -417,10 +417,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX1-NEXT: vpsraw $4, %xmm4, %xmm5 ; AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm4 ; AVX1-NEXT: vpsraw $2, %xmm4, %xmm5 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm4 +; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm6 +; AVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm4, %xmm4 ; AVX1-NEXT: vpsraw $1, %xmm4, %xmm5 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 +; AVX1-NEXT: vpsllw $2, %xmm3, %xmm3 ; AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm3 ; AVX1-NEXT: vpsrlw $8, %xmm3, %xmm3 ; AVX1-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -428,10 +428,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX1-NEXT: vpsraw $4, %xmm0, %xmm4 ; AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 ; AVX1-NEXT: vpsraw $2, %xmm0, %xmm4 -; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 +; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm5 +; AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm0, %xmm0 ; AVX1-NEXT: vpsraw $1, %xmm0, %xmm4 -; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm1 ; AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 ; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm0 ; AVX1-NEXT: vpackuswb %xmm3, %xmm0, %xmm0 @@ -446,10 +446,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX2-NEXT: vpsraw $4, %ymm3, %ymm4 ; AVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; AVX2-NEXT: vpsraw $2, %ymm3, %ymm4 -; AVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; AVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; AVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; AVX2-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; AVX2-NEXT: vpsraw $1, %ymm3, %ymm4 -; AVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; AVX2-NEXT: vpsllw $2, %ymm2, %ymm2 ; AVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; AVX2-NEXT: vpsrlw $8, %ymm2, %ymm2 ; AVX2-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -457,10 +457,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX2-NEXT: vpsraw $4, %ymm0, %ymm3 ; AVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX2-NEXT: vpsraw $2, %ymm0, %ymm3 -; AVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; AVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; AVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; AVX2-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; AVX2-NEXT: vpsraw $1, %ymm0, %ymm3 -; AVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; AVX2-NEXT: vpsllw $2, %ymm1, %ymm1 ; AVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX2-NEXT: vpsrlw $8, %ymm0, %ymm0 ; AVX2-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 @@ -498,10 +498,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX512DQ-NEXT: vpsraw $4, %ymm3, %ymm4 ; AVX512DQ-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; AVX512DQ-NEXT: vpsraw $2, %ymm3, %ymm4 -; AVX512DQ-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; AVX512DQ-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; AVX512DQ-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; AVX512DQ-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; AVX512DQ-NEXT: vpsraw $1, %ymm3, %ymm4 -; AVX512DQ-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; AVX512DQ-NEXT: vpsllw $2, %ymm2, %ymm2 ; AVX512DQ-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; AVX512DQ-NEXT: vpsrlw $8, %ymm2, %ymm2 ; AVX512DQ-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -509,10 +509,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX512DQ-NEXT: vpsraw $4, %ymm0, %ymm3 ; AVX512DQ-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpsraw $2, %ymm0, %ymm3 -; AVX512DQ-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; AVX512DQ-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; AVX512DQ-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; AVX512DQ-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpsraw $1, %ymm0, %ymm3 -; AVX512DQ-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; AVX512DQ-NEXT: vpsllw $2, %ymm1, %ymm1 ; AVX512DQ-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpsrlw $8, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 @@ -534,10 +534,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX512DQVL-NEXT: vpsraw $4, %ymm3, %ymm4 ; AVX512DQVL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; AVX512DQVL-NEXT: vpsraw $2, %ymm3, %ymm4 -; AVX512DQVL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; AVX512DQVL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; AVX512DQVL-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; AVX512DQVL-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; AVX512DQVL-NEXT: vpsraw $1, %ymm3, %ymm4 -; AVX512DQVL-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; AVX512DQVL-NEXT: vpsllw $2, %ymm2, %ymm2 ; AVX512DQVL-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; AVX512DQVL-NEXT: vpsrlw $8, %ymm2, %ymm2 ; AVX512DQVL-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -545,10 +545,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; AVX512DQVL-NEXT: vpsraw $4, %ymm0, %ymm3 ; AVX512DQVL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX512DQVL-NEXT: vpsraw $2, %ymm0, %ymm3 -; AVX512DQVL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; AVX512DQVL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; AVX512DQVL-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; AVX512DQVL-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; AVX512DQVL-NEXT: vpsraw $1, %ymm0, %ymm3 -; AVX512DQVL-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; AVX512DQVL-NEXT: vpsllw $2, %ymm1, %ymm1 ; AVX512DQVL-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; AVX512DQVL-NEXT: vpsrlw $8, %ymm0, %ymm0 ; AVX512DQVL-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 @@ -572,10 +572,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; X86-AVX1-NEXT: vpsraw $4, %xmm5, %xmm6 ; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm5 ; X86-AVX1-NEXT: vpsraw $2, %xmm5, %xmm6 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm5 +; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm7 +; X86-AVX1-NEXT: vpblendvb %xmm7, %xmm6, %xmm5, %xmm5 ; X86-AVX1-NEXT: vpsraw $1, %xmm5, %xmm6 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 +; X86-AVX1-NEXT: vpsllw $2, %xmm3, %xmm3 ; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm6, %xmm5, %xmm3 ; X86-AVX1-NEXT: vpsrlw $8, %xmm3, %xmm3 ; X86-AVX1-NEXT: vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -583,10 +583,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; X86-AVX1-NEXT: vpsraw $4, %xmm4, %xmm5 ; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 ; X86-AVX1-NEXT: vpsraw $2, %xmm4, %xmm5 -; X86-AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 +; X86-AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm6 +; X86-AVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm4, %xmm4 ; X86-AVX1-NEXT: vpsraw $1, %xmm4, %xmm5 -; X86-AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; X86-AVX1-NEXT: vpsllw $2, %xmm2, %xmm2 ; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm2 ; X86-AVX1-NEXT: vpsrlw $8, %xmm2, %xmm2 ; X86-AVX1-NEXT: vpackuswb %xmm3, %xmm2, %xmm2 @@ -596,10 +596,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; X86-AVX1-NEXT: vpsraw $4, %xmm4, %xmm5 ; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm4 ; X86-AVX1-NEXT: vpsraw $2, %xmm4, %xmm5 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm4 +; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm6 +; X86-AVX1-NEXT: vpblendvb %xmm6, %xmm5, %xmm4, %xmm4 ; X86-AVX1-NEXT: vpsraw $1, %xmm4, %xmm5 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 +; X86-AVX1-NEXT: vpsllw $2, %xmm3, %xmm3 ; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm3 ; X86-AVX1-NEXT: vpsrlw $8, %xmm3, %xmm3 ; X86-AVX1-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -607,10 +607,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; X86-AVX1-NEXT: vpsraw $4, %xmm0, %xmm4 ; X86-AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 ; X86-AVX1-NEXT: vpsraw $2, %xmm0, %xmm4 -; X86-AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; X86-AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 +; X86-AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm5 +; X86-AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm0, %xmm0 ; X86-AVX1-NEXT: vpsraw $1, %xmm0, %xmm4 -; X86-AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; X86-AVX1-NEXT: vpsllw $2, %xmm1, %xmm1 ; X86-AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 ; X86-AVX1-NEXT: vpsrlw $8, %xmm0, %xmm0 ; X86-AVX1-NEXT: vpackuswb %xmm3, %xmm0, %xmm0 @@ -625,10 +625,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; X86-AVX2-NEXT: vpsraw $4, %ymm3, %ymm4 ; X86-AVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 ; X86-AVX2-NEXT: vpsraw $2, %ymm3, %ymm4 -; X86-AVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; X86-AVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm3 +; X86-AVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm5 +; X86-AVX2-NEXT: vpblendvb %ymm5, %ymm4, %ymm3, %ymm3 ; X86-AVX2-NEXT: vpsraw $1, %ymm3, %ymm4 -; X86-AVX2-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; X86-AVX2-NEXT: vpsllw $2, %ymm2, %ymm2 ; X86-AVX2-NEXT: vpblendvb %ymm2, %ymm4, %ymm3, %ymm2 ; X86-AVX2-NEXT: vpsrlw $8, %ymm2, %ymm2 ; X86-AVX2-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -636,10 +636,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind { ; X86-AVX2-NEXT: vpsraw $4, %ymm0, %ymm3 ; X86-AVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; X86-AVX2-NEXT: vpsraw $2, %ymm0, %ymm3 -; X86-AVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; X86-AVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 +; X86-AVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm4 +; X86-AVX2-NEXT: vpblendvb %ymm4, %ymm3, %ymm0, %ymm0 ; X86-AVX2-NEXT: vpsraw $1, %ymm0, %ymm3 -; X86-AVX2-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; X86-AVX2-NEXT: vpsllw $2, %ymm1, %ymm1 ; X86-AVX2-NEXT: vpblendvb %ymm1, %ymm3, %ymm0, %ymm0 ; X86-AVX2-NEXT: vpsrlw $8, %ymm0, %ymm0 ; X86-AVX2-NEXT: vpackuswb %ymm2, %ymm0, %ymm0 diff --git a/llvm/test/CodeGen/X86/vector-shift-ashr-512.ll b/llvm/test/CodeGen/X86/vector-shift-ashr-512.ll index ea0745b..0fb0420 100644 --- a/llvm/test/CodeGen/X86/vector-shift-ashr-512.ll +++ b/llvm/test/CodeGen/X86/vector-shift-ashr-512.ll @@ -59,10 +59,10 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; AVX512DQ-NEXT: vpsraw $4, %ymm5, %ymm6 ; AVX512DQ-NEXT: vpblendvb %ymm3, %ymm6, %ymm5, %ymm5 ; AVX512DQ-NEXT: vpsraw $2, %ymm5, %ymm6 -; AVX512DQ-NEXT: vpaddw %ymm3, %ymm3, %ymm3 -; AVX512DQ-NEXT: vpblendvb %ymm3, %ymm6, %ymm5, %ymm5 +; AVX512DQ-NEXT: vpaddw %ymm3, %ymm3, %ymm7 +; AVX512DQ-NEXT: vpblendvb %ymm7, %ymm6, %ymm5, %ymm5 ; AVX512DQ-NEXT: vpsraw $1, %ymm5, %ymm6 -; AVX512DQ-NEXT: vpaddw %ymm3, %ymm3, %ymm3 +; AVX512DQ-NEXT: vpsllw $2, %ymm3, %ymm3 ; AVX512DQ-NEXT: vpblendvb %ymm3, %ymm6, %ymm5, %ymm3 ; AVX512DQ-NEXT: vpsrlw $8, %ymm3, %ymm3 ; AVX512DQ-NEXT: vpunpcklbw {{.*#+}} ymm2 = ymm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -70,10 +70,10 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; AVX512DQ-NEXT: vpsraw $4, %ymm4, %ymm5 ; AVX512DQ-NEXT: vpblendvb %ymm2, %ymm5, %ymm4, %ymm4 ; AVX512DQ-NEXT: vpsraw $2, %ymm4, %ymm5 -; AVX512DQ-NEXT: vpaddw %ymm2, %ymm2, %ymm2 -; AVX512DQ-NEXT: vpblendvb %ymm2, %ymm5, %ymm4, %ymm4 +; AVX512DQ-NEXT: vpaddw %ymm2, %ymm2, %ymm6 +; AVX512DQ-NEXT: vpblendvb %ymm6, %ymm5, %ymm4, %ymm4 ; AVX512DQ-NEXT: vpsraw $1, %ymm4, %ymm5 -; AVX512DQ-NEXT: vpaddw %ymm2, %ymm2, %ymm2 +; AVX512DQ-NEXT: vpsllw $2, %ymm2, %ymm2 ; AVX512DQ-NEXT: vpblendvb %ymm2, %ymm5, %ymm4, %ymm2 ; AVX512DQ-NEXT: vpsrlw $8, %ymm2, %ymm2 ; AVX512DQ-NEXT: vpackuswb %ymm3, %ymm2, %ymm2 @@ -83,10 +83,10 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; AVX512DQ-NEXT: vpsraw $4, %ymm4, %ymm5 ; AVX512DQ-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm4 ; AVX512DQ-NEXT: vpsraw $2, %ymm4, %ymm5 -; AVX512DQ-NEXT: vpaddw %ymm3, %ymm3, %ymm3 -; AVX512DQ-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm4 +; AVX512DQ-NEXT: vpaddw %ymm3, %ymm3, %ymm6 +; AVX512DQ-NEXT: vpblendvb %ymm6, %ymm5, %ymm4, %ymm4 ; AVX512DQ-NEXT: vpsraw $1, %ymm4, %ymm5 -; AVX512DQ-NEXT: vpaddw %ymm3, %ymm3, %ymm3 +; AVX512DQ-NEXT: vpsllw $2, %ymm3, %ymm3 ; AVX512DQ-NEXT: vpblendvb %ymm3, %ymm5, %ymm4, %ymm3 ; AVX512DQ-NEXT: vpsrlw $8, %ymm3, %ymm3 ; AVX512DQ-NEXT: vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23] @@ -94,10 +94,10 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; AVX512DQ-NEXT: vpsraw $4, %ymm0, %ymm4 ; AVX512DQ-NEXT: vpblendvb %ymm1, %ymm4, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpsraw $2, %ymm0, %ymm4 -; AVX512DQ-NEXT: vpaddw %ymm1, %ymm1, %ymm1 -; AVX512DQ-NEXT: vpblendvb %ymm1, %ymm4, %ymm0, %ymm0 +; AVX512DQ-NEXT: vpaddw %ymm1, %ymm1, %ymm5 +; AVX512DQ-NEXT: vpblendvb %ymm5, %ymm4, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpsraw $1, %ymm0, %ymm4 -; AVX512DQ-NEXT: vpaddw %ymm1, %ymm1, %ymm1 +; AVX512DQ-NEXT: vpsllw $2, %ymm1, %ymm1 ; AVX512DQ-NEXT: vpblendvb %ymm1, %ymm4, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpsrlw $8, %ymm0, %ymm0 ; AVX512DQ-NEXT: vpackuswb %ymm3, %ymm0, %ymm0 @@ -113,11 +113,11 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; AVX512BW-NEXT: vpmovb2m %zmm4, %k1 ; AVX512BW-NEXT: vmovdqu8 %zmm3, %zmm2 {%k1} ; AVX512BW-NEXT: vpsraw $2, %zmm2, %zmm3 -; AVX512BW-NEXT: vpaddw %zmm4, %zmm4, %zmm4 -; AVX512BW-NEXT: vpmovb2m %zmm4, %k1 +; AVX512BW-NEXT: vpaddw %zmm4, %zmm4, %zmm5 +; AVX512BW-NEXT: vpmovb2m %zmm5, %k1 ; AVX512BW-NEXT: vmovdqu8 %zmm3, %zmm2 {%k1} ; AVX512BW-NEXT: vpsraw $1, %zmm2, %zmm3 -; AVX512BW-NEXT: vpaddw %zmm4, %zmm4, %zmm4 +; AVX512BW-NEXT: vpsllw $2, %zmm4, %zmm4 ; AVX512BW-NEXT: vpmovb2m %zmm4, %k1 ; AVX512BW-NEXT: vmovdqu8 %zmm3, %zmm2 {%k1} ; AVX512BW-NEXT: vpsrlw $8, %zmm2, %zmm2 @@ -127,11 +127,11 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) nounwind { ; AVX512BW-NEXT: vpmovb2m %zmm1, %k1 ; AVX512BW-NEXT: vmovdqu8 %zmm3, %zmm0 {%k1} ; AVX512BW-NEXT: vpsraw $2, %zmm0, %zmm3 -; AVX512BW-NEXT: vpaddw %zmm1, %zmm1, %zmm1 -; AVX512BW-NEXT: vpmovb2m %zmm1, %k1 +; AVX512BW-NEXT: vpaddw %zmm1, %zmm1, %zmm4 +; AVX512BW-NEXT: vpmovb2m %zmm4, %k1 ; AVX512BW-NEXT: vmovdqu8 %zmm3, %zmm0 {%k1} ; AVX512BW-NEXT: vpsraw $1, %zmm0, %zmm3 -; AVX512BW-NEXT: vpaddw %zmm1, %zmm1, %zmm1 +; AVX512BW-NEXT: vpsllw $2, %zmm1, %zmm1 ; AVX512BW-NEXT: vpmovb2m %zmm1, %k1 ; AVX512BW-NEXT: vmovdqu8 %zmm3, %zmm0 {%k1} ; AVX512BW-NEXT: vpsrlw $8, %zmm0, %zmm0 diff --git a/llvm/test/CodeGen/X86/vector-shift-ashr-sub128.ll b/llvm/test/CodeGen/X86/vector-shift-ashr-sub128.ll index f7de8d4..c5d3297 100644 --- a/llvm/test/CodeGen/X86/vector-shift-ashr-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-shift-ashr-sub128.ll @@ -196,14 +196,14 @@ define <4 x i16> @var_shift_v4i16(<4 x i16> %a, <4 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm2 ; AVX1-NEXT: vpsraw $8, %xmm0, %xmm3 ; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $4, %xmm0, %xmm3 +; AVX1-NEXT: vpblendvb %xmm2, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $2, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm3 +; AVX1-NEXT: vpblendvb %xmm3, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $1, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: retq ; ; AVX2-LABEL: var_shift_v4i16: @@ -367,14 +367,14 @@ define <2 x i16> @var_shift_v2i16(<2 x i16> %a, <2 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm2 ; AVX1-NEXT: vpsraw $8, %xmm0, %xmm3 ; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsraw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $4, %xmm0, %xmm3 +; AVX1-NEXT: vpblendvb %xmm2, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $2, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm3 +; AVX1-NEXT: vpblendvb %xmm3, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $1, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: retq ; ; AVX2-LABEL: var_shift_v2i16: @@ -568,10 +568,10 @@ define <8 x i8> @var_shift_v8i8(<8 x i8> %a, <8 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm3, %xmm4 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $2, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 +; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; AVX-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $1, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; AVX-NEXT: vpsllw $2, %xmm2, %xmm2 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; AVX-NEXT: vpsrlw $8, %xmm2, %xmm2 ; AVX-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -579,10 +579,10 @@ define <8 x i8> @var_shift_v8i8(<8 x i8> %a, <8 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm0, %xmm3 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $2, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 +; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm4 +; AVX-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $1, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpsllw $2, %xmm1, %xmm1 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0 ; AVX-NEXT: vpackuswb %xmm2, %xmm0, %xmm0 @@ -796,10 +796,10 @@ define <4 x i8> @var_shift_v4i8(<4 x i8> %a, <4 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm3, %xmm4 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $2, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 +; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; AVX-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $1, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; AVX-NEXT: vpsllw $2, %xmm2, %xmm2 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; AVX-NEXT: vpsrlw $8, %xmm2, %xmm2 ; AVX-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -807,10 +807,10 @@ define <4 x i8> @var_shift_v4i8(<4 x i8> %a, <4 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm0, %xmm3 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $2, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 +; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm4 +; AVX-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $1, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpsllw $2, %xmm1, %xmm1 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0 ; AVX-NEXT: vpackuswb %xmm2, %xmm0, %xmm0 @@ -1024,10 +1024,10 @@ define <2 x i8> @var_shift_v2i8(<2 x i8> %a, <2 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm3, %xmm4 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $2, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm3 +; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm5 +; AVX-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 ; AVX-NEXT: vpsraw $1, %xmm3, %xmm4 -; AVX-NEXT: vpaddw %xmm2, %xmm2, %xmm2 +; AVX-NEXT: vpsllw $2, %xmm2, %xmm2 ; AVX-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; AVX-NEXT: vpsrlw $8, %xmm2, %xmm2 ; AVX-NEXT: vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] @@ -1035,10 +1035,10 @@ define <2 x i8> @var_shift_v2i8(<2 x i8> %a, <2 x i8> %b) nounwind { ; AVX-NEXT: vpsraw $4, %xmm0, %xmm3 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $2, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 -; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 +; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm4 +; AVX-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsraw $1, %xmm0, %xmm3 -; AVX-NEXT: vpaddw %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpsllw $2, %xmm1, %xmm1 ; AVX-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0 ; AVX-NEXT: vpackuswb %xmm2, %xmm0, %xmm0 diff --git a/llvm/test/CodeGen/X86/vector-shift-lshr-128.ll b/llvm/test/CodeGen/X86/vector-shift-lshr-128.ll index 1d1697a..8cb2c7b 100644 --- a/llvm/test/CodeGen/X86/vector-shift-lshr-128.ll +++ b/llvm/test/CodeGen/X86/vector-shift-lshr-128.ll @@ -262,14 +262,14 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm2 ; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm3 ; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm3 +; AVX1-NEXT: vpblendvb %xmm2, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm3 +; AVX1-NEXT: vpblendvb %xmm3, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: retq ; ; AVX2-LABEL: var_shift_v8i16: diff --git a/llvm/test/CodeGen/X86/vector-shift-lshr-256.ll b/llvm/test/CodeGen/X86/vector-shift-lshr-256.ll index 3a4bb22..606adb4 100644 --- a/llvm/test/CodeGen/X86/vector-shift-lshr-256.ll +++ b/llvm/test/CodeGen/X86/vector-shift-lshr-256.ll @@ -198,29 +198,29 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm3 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm4 ; AVX1-NEXT: vpsrlw $8, %xmm4, %xmm5 -; AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm2 -; AVX1-NEXT: vpsrlw $4, %xmm2, %xmm4 -; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; AVX1-NEXT: vpsrlw $2, %xmm2, %xmm4 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; AVX1-NEXT: vpsrlw $1, %xmm2, %xmm4 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 +; AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 +; AVX1-NEXT: vpsrlw $4, %xmm4, %xmm5 +; AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm3 +; AVX1-NEXT: vpsrlw $2, %xmm3, %xmm4 +; AVX1-NEXT: vpsllw $2, %xmm2, %xmm5 +; AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vpsrlw $1, %xmm3, %xmm4 +; AVX1-NEXT: vpsllw $3, %xmm2, %xmm2 +; AVX1-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; AVX1-NEXT: vpsllw $12, %xmm1, %xmm3 ; AVX1-NEXT: vpsllw $4, %xmm1, %xmm1 ; AVX1-NEXT: vpor %xmm3, %xmm1, %xmm1 ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm3 ; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm4 ; AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm4 +; AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm3 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm4 +; AVX1-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm3 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: retq ; @@ -300,29 +300,29 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) nounwind { ; X86-AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm3 ; X86-AVX1-NEXT: vextractf128 $1, %ymm0, %xmm4 ; X86-AVX1-NEXT: vpsrlw $8, %xmm4, %xmm5 -; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm2 -; X86-AVX1-NEXT: vpsrlw $4, %xmm2, %xmm4 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; X86-AVX1-NEXT: vpsrlw $2, %xmm2, %xmm4 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 -; X86-AVX1-NEXT: vpsrlw $1, %xmm2, %xmm4 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm2, %xmm2 +; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm5, %xmm4, %xmm4 +; X86-AVX1-NEXT: vpsrlw $4, %xmm4, %xmm5 +; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm5, %xmm4, %xmm3 +; X86-AVX1-NEXT: vpsrlw $2, %xmm3, %xmm4 +; X86-AVX1-NEXT: vpsllw $2, %xmm2, %xmm5 +; X86-AVX1-NEXT: vpblendvb %xmm5, %xmm4, %xmm3, %xmm3 +; X86-AVX1-NEXT: vpsrlw $1, %xmm3, %xmm4 +; X86-AVX1-NEXT: vpsllw $3, %xmm2, %xmm2 +; X86-AVX1-NEXT: vpblendvb %xmm2, %xmm4, %xmm3, %xmm2 ; X86-AVX1-NEXT: vpsllw $12, %xmm1, %xmm3 ; X86-AVX1-NEXT: vpsllw $4, %xmm1, %xmm1 ; X86-AVX1-NEXT: vpor %xmm3, %xmm1, %xmm1 ; X86-AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm3 ; X86-AVX1-NEXT: vpsrlw $8, %xmm0, %xmm4 ; X86-AVX1-NEXT: vpblendvb %xmm1, %xmm4, %xmm0, %xmm0 -; X86-AVX1-NEXT: vpsrlw $4, %xmm0, %xmm1 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; X86-AVX1-NEXT: vpsrlw $2, %xmm0, %xmm1 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 -; X86-AVX1-NEXT: vpsrlw $1, %xmm0, %xmm1 -; X86-AVX1-NEXT: vpaddw %xmm3, %xmm3, %xmm3 -; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 +; X86-AVX1-NEXT: vpsrlw $4, %xmm0, %xmm4 +; X86-AVX1-NEXT: vpblendvb %xmm3, %xmm4, %xmm0, %xmm0 +; X86-AVX1-NEXT: vpsrlw $2, %xmm0, %xmm3 +; X86-AVX1-NEXT: vpsllw $2, %xmm1, %xmm4 +; X86-AVX1-NEXT: vpblendvb %xmm4, %xmm3, %xmm0, %xmm0 +; X86-AVX1-NEXT: vpsrlw $1, %xmm0, %xmm3 +; X86-AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; X86-AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 ; X86-AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; X86-AVX1-NEXT: retl ; diff --git a/llvm/test/CodeGen/X86/vector-shift-lshr-sub128.ll b/llvm/test/CodeGen/X86/vector-shift-lshr-sub128.ll index 7928111..57874c4 100644 --- a/llvm/test/CodeGen/X86/vector-shift-lshr-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-shift-lshr-sub128.ll @@ -196,14 +196,14 @@ define <4 x i16> @var_shift_v4i16(<4 x i16> %a, <4 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm2 ; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm3 ; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm3 +; AVX1-NEXT: vpblendvb %xmm2, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm3 +; AVX1-NEXT: vpblendvb %xmm3, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: retq ; ; AVX2-LABEL: var_shift_v4i16: @@ -367,14 +367,14 @@ define <2 x i16> @var_shift_v2i16(<2 x i16> %a, <2 x i16> %b) nounwind { ; AVX1-NEXT: vpaddw %xmm1, %xmm1, %xmm2 ; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm3 ; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm3 +; AVX1-NEXT: vpblendvb %xmm2, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm3 +; AVX1-NEXT: vpblendvb %xmm3, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: retq ; ; AVX2-LABEL: var_shift_v2i16: diff --git a/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll b/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll index dbbfaab..be41945 100644 --- a/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll +++ b/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll @@ -8079,14 +8079,14 @@ define <16 x i16> @pr43230(<16 x i16> %a, <16 x i16> %b) { ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0 ; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm3 ; AVX1-NEXT: vpblendvb %xmm1, %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm1 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm1 -; AVX1-NEXT: vpaddw %xmm2, %xmm2, %xmm2 -; AVX1-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm3 +; AVX1-NEXT: vpblendvb %xmm2, %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $2, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $2, %xmm1, %xmm3 +; AVX1-NEXT: vpblendvb %xmm3, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpsrlw $1, %xmm0, %xmm2 +; AVX1-NEXT: vpsllw $3, %xmm1, %xmm1 +; AVX1-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0 ; AVX1-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0 ; AVX1-NEXT: retq diff --git a/llvm/test/CodeGen/X86/vector-zext.ll b/llvm/test/CodeGen/X86/vector-zext.ll index bd1a48b..7b0f1c9 100644 --- a/llvm/test/CodeGen/X86/vector-zext.ll +++ b/llvm/test/CodeGen/X86/vector-zext.ll @@ -2555,7 +2555,7 @@ entry: define <4 x i64> @splatshuf_zext_v4i64(<4 x i32> %x) { ; SSE2-LABEL: splatshuf_zext_v4i64: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: pxor %xmm1, %xmm1 ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: movdqa %xmm0, %xmm1 @@ -2563,7 +2563,7 @@ define <4 x i64> @splatshuf_zext_v4i64(<4 x i32> %x) { ; ; SSSE3-LABEL: splatshuf_zext_v4i64: ; SSSE3: # %bb.0: -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSSE3-NEXT: pxor %xmm1, %xmm1 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSSE3-NEXT: movdqa %xmm0, %xmm1 @@ -2571,7 +2571,7 @@ define <4 x i64> @splatshuf_zext_v4i64(<4 x i32> %x) { ; ; SSE41-LABEL: splatshuf_zext_v4i64: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE41-NEXT: pmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero ; SSE41-NEXT: movdqa %xmm0, %xmm1 ; SSE41-NEXT: retq diff --git a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll index 19a31a6..31ed745 100644 --- a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll +++ b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll @@ -911,7 +911,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 16(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: paddb (%rdx), %xmm0 ; SSE2-NEXT: movdqa %xmm0, (%rcx) @@ -1898,7 +1898,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 32(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: movdqa %xmm0, %xmm3 ; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3] @@ -4610,7 +4610,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 48(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: movdqa %xmm0, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE2-NEXT: pxor %xmm1, %xmm1 @@ -6544,7 +6544,7 @@ define void @vec512_i32_widen_to_i64_factor2_broadcast_to_v8i64_factor8(ptr %in. ; SSE2-NEXT: movdqa (%rdi), %xmm0 ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: pxor %xmm1, %xmm1 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: movdqa 16(%rdx), %xmm1 ; SSE2-NEXT: paddb %xmm0, %xmm1 diff --git a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll index 239472c..5b4cdd2 100644 --- a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll +++ b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll @@ -769,7 +769,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-LABEL: vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] ; SSE2-NEXT: paddb (%rsi), %xmm1 ; SSE2-NEXT: movdqa %xmm1, (%rdx) @@ -1522,7 +1522,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-LABEL: vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: movdqa %xmm1, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] @@ -3660,7 +3660,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2-LABEL: vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: movdqa %xmm1, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1] ; SSE2-NEXT: pxor %xmm0, %xmm0 @@ -5250,7 +5250,7 @@ define void @vec512_i16_widen_to_i256_factor16_broadcast_to_v2i256_factor2(ptr % define void @vec512_i32_widen_to_i64_factor2_broadcast_to_v8i64_factor8(ptr %in.elt.ptr, ptr %out.vec.bias.ptr, ptr %out.vec.ptr) nounwind { ; SSE2-LABEL: vec512_i32_widen_to_i64_factor2_broadcast_to_v8i64_factor8: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0] ; SSE2-NEXT: pxor %xmm1, %xmm1 ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: movdqa 16(%rsi), %xmm1 diff --git a/llvm/test/MC/AMDGPU/hsa-gfx1250-v4.s b/llvm/test/MC/AMDGPU/hsa-gfx1250-v4.s index 3c69361..80a340c 100644 --- a/llvm/test/MC/AMDGPU/hsa-gfx1250-v4.s +++ b/llvm/test/MC/AMDGPU/hsa-gfx1250-v4.s @@ -178,6 +178,7 @@ max_vgprs: // ASM-NEXT: .amdhsa_next_free_sgpr 32 // ASM-NEXT: .amdhsa_named_barrier_count 3 // ASM-NEXT: .amdhsa_reserve_vcc 0 +// ASM-NEXT: .amdhsa_reserve_xnack_mask 1 // ASM-NEXT: .amdhsa_float_round_mode_32 1 // ASM-NEXT: .amdhsa_float_round_mode_16_64 1 // ASM-NEXT: .amdhsa_float_denorm_mode_32 1 diff --git a/llvm/test/MC/AMDGPU/hsa-gfx1251-v4.s b/llvm/test/MC/AMDGPU/hsa-gfx1251-v4.s index 776006b..642e62d 100644 --- a/llvm/test/MC/AMDGPU/hsa-gfx1251-v4.s +++ b/llvm/test/MC/AMDGPU/hsa-gfx1251-v4.s @@ -178,6 +178,7 @@ max_vgprs: // ASM-NEXT: .amdhsa_next_free_sgpr 32 // ASM-NEXT: .amdhsa_named_barrier_count 3 // ASM-NEXT: .amdhsa_reserve_vcc 0 +// ASM-NEXT: .amdhsa_reserve_xnack_mask 1 // ASM-NEXT: .amdhsa_float_round_mode_32 1 // ASM-NEXT: .amdhsa_float_round_mode_16_64 1 // ASM-NEXT: .amdhsa_float_denorm_mode_32 1 diff --git a/llvm/test/TableGen/RegisterClassCopyCost.td b/llvm/test/TableGen/RegisterClassCopyCost.td new file mode 100644 index 0000000..fc65fdb --- /dev/null +++ b/llvm/test/TableGen/RegisterClassCopyCost.td @@ -0,0 +1,31 @@ +// RUN: llvm-tblgen --gen-register-info -I %p/../../include %s 2>&1 | FileCheck %s +// RUN: not llvm-tblgen --gen-register-info -I %p/../../include -DERROR %s 2>&1 | FileCheck -check-prefix=ERROR %s + +// Check that there is no assertion when specifying unsupported +// CopyCost values on register classes. Check that negative CopyCost +// values are saturated to 255. + +include "llvm/Target/Target.td" + +// CHECK: extern const MCRegisterClass MyTargetMCRegisterClasses[] = { +// CHECK-NEXT: { GPR32, GPR32Bits, 0, 2, sizeof(GPR32Bits), MyTarget::GPR32RegClassID, 32, 1, true, false }, +// CHECK-NEXT: { SPECIAL_CLASS, SPECIAL_CLASSBits, 6, 1, sizeof(SPECIAL_CLASSBits), MyTarget::SPECIAL_CLASSRegClassID, 32, 255, true, false }, +// CHECK-NEXT: }; + +def MyTargetISA : InstrInfo; +def MyTarget : Target { let InstructionSet = MyTargetISA; } + +def R0 : Register<"r0"> { let Namespace = "MyTarget"; } +def R1 : Register<"r1"> { let Namespace = "MyTarget"; } +def SPECIAL : Register<"special"> { let Namespace = "MyTarget"; } + +// ERROR: :[[@LINE+1]]:5: error: 'CopyCost' must be an 8-bit value +def GPR32 : RegisterClass<"MyTarget", [i32], 32, (add R0, R1)> { +#ifdef ERROR + let CopyCost = 256; +#endif +} + +def SPECIAL_CLASS : RegisterClass<"MyTarget", [i32], 32, (add SPECIAL)> { + let CopyCost = -1; +} diff --git a/llvm/test/Transforms/InstCombine/masked_intrinsics.ll b/llvm/test/Transforms/InstCombine/masked_intrinsics.ll index 8f76834..67ab167 100644 --- a/llvm/test/Transforms/InstCombine/masked_intrinsics.ll +++ b/llvm/test/Transforms/InstCombine/masked_intrinsics.ll @@ -16,6 +16,14 @@ define <2 x double> @load_zeromask(ptr %ptr, <2 x double> %passthru) { ret <2 x double> %res } +define <2 x double> @load_zero_withpoison_mask(ptr %ptr, <2 x double> %passthru) { +; CHECK-LABEL: @load_zero_withpoison_mask( +; CHECK-NEXT: ret <2 x double> [[PASSTHRU:%.*]] +; + %res = call <2 x double> @llvm.masked.load.v2f64.p0(ptr %ptr, i32 1, <2 x i1> <i1 0, i1 poison>, <2 x double> %passthru) + ret <2 x double> %res +} + define <2 x double> @load_onemask(ptr %ptr, <2 x double> %passthru) { ; CHECK-LABEL: @load_onemask( ; CHECK-NEXT: [[UNMASKEDLOAD:%.*]] = load <2 x double>, ptr [[PTR:%.*]], align 2 @@ -150,6 +158,14 @@ define void @store_zeromask(ptr %ptr, <2 x double> %val) { ret void } +define void @store_poisonmask(ptr %ptr, <2 x double> %val) { +; CHECK-LABEL: @store_poisonmask( +; CHECK-NEXT: ret void +; + call void @llvm.masked.store.v2f64.p0(<2 x double> %val, ptr %ptr, i32 4, <2 x i1> splat(i1 poison)) + ret void +} + define void @store_onemask(ptr %ptr, <2 x double> %val) { ; CHECK-LABEL: @store_onemask( ; CHECK-NEXT: store <2 x double> [[VAL:%.*]], ptr [[PTR:%.*]], align 4 @@ -159,6 +175,15 @@ define void @store_onemask(ptr %ptr, <2 x double> %val) { ret void } +define void @store_one_withpoison_mask(ptr %ptr, <2 x double> %val) { +; CHECK-LABEL: @store_one_withpoison_mask( +; CHECK-NEXT: store <2 x double> [[VAL:%.*]], ptr [[PTR:%.*]], align 4 +; CHECK-NEXT: ret void +; + call void @llvm.masked.store.v2f64.p0(<2 x double> %val, ptr %ptr, i32 4, <2 x i1> <i1 1, i1 poison>) + ret void +} + define void @store_demandedelts(ptr %ptr, double %val) { ; CHECK-LABEL: @store_demandedelts( ; CHECK-NEXT: [[VALVEC1:%.*]] = insertelement <2 x double> poison, double [[VAL:%.*]], i64 0 @@ -189,6 +214,13 @@ define <2 x double> @gather_zeromask(<2 x ptr> %ptrs, <2 x double> %passthru) { ret <2 x double> %res } +define <2 x double> @gather_zero_withpoison_mask(<2 x ptr> %ptrs, <2 x double> %passthru) { +; CHECK-LABEL: @gather_zero_withpoison_mask( +; CHECK-NEXT: ret <2 x double> [[PASSTHRU:%.*]] +; + %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> %ptrs, i32 4, <2 x i1> <i1 0, i1 poison>, <2 x double> %passthru) + ret <2 x double> %res +} define <2 x double> @gather_onemask(<2 x ptr> %ptrs, <2 x double> %passthru) { ; CHECK-LABEL: @gather_onemask( @@ -199,6 +231,15 @@ define <2 x double> @gather_onemask(<2 x ptr> %ptrs, <2 x double> %passthru) { ret <2 x double> %res } +define <2 x double> @gather_one_withpoisonmask(<2 x ptr> %ptrs, <2 x double> %passthru) { +; CHECK-LABEL: @gather_one_withpoisonmask( +; CHECK-NEXT: [[RES:%.*]] = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> [[PTRS:%.*]], i32 4, <2 x i1> <i1 true, i1 poison>, <2 x double> [[PASSTHRU:%.*]]) +; CHECK-NEXT: ret <2 x double> [[RES]] +; + %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> %ptrs, i32 4, <2 x i1> <i1 true, i1 poison>, <2 x double> %passthru) + ret <2 x double> %res +} + define <4 x double> @gather_lane2(ptr %base, double %pt) { ; CHECK-LABEL: @gather_lane2( ; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <4 x i64> <i64 poison, i64 poison, i64 2, i64 poison> @@ -257,6 +298,23 @@ define void @scatter_zeromask(<2 x ptr> %ptrs, <2 x double> %val) { ret void } +define void @scatter_zero_withpoison_mask(<2 x ptr> %ptrs, <2 x double> %val) { +; CHECK-LABEL: @scatter_zero_withpoison_mask( +; CHECK-NEXT: ret void +; + call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> %val, <2 x ptr> %ptrs, i32 8, <2 x i1> <i1 0, i1 poison>) + ret void +} + +define void @scatter_one_withpoison_mask(<2 x ptr> %ptrs, <2 x double> %val) { +; CHECK-LABEL: @scatter_one_withpoison_mask( +; CHECK-NEXT: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> [[VAL:%.*]], <2 x ptr> [[PTRS:%.*]], i32 8, <2 x i1> <i1 true, i1 poison>) +; CHECK-NEXT: ret void +; + call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> %val, <2 x ptr> %ptrs, i32 8, <2 x i1> <i1 1, i1 poison>) + ret void +} + define void @scatter_demandedelts(ptr %ptr, double %val) { ; CHECK-LABEL: @scatter_demandedelts( ; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 poison> diff --git a/llvm/test/Transforms/InstCombine/pr83947.ll b/llvm/test/Transforms/InstCombine/pr83947.ll index 1906502..679230a4 100644 --- a/llvm/test/Transforms/InstCombine/pr83947.ll +++ b/llvm/test/Transforms/InstCombine/pr83947.ll @@ -24,7 +24,6 @@ define void @masked_scatter2() { define void @masked_scatter3() { ; CHECK-LABEL: define void @masked_scatter3() { -; CHECK-NEXT: store i32 0, ptr @c, align 4 ; CHECK-NEXT: ret void ; call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> zeroinitializer, <2 x ptr> splat (ptr @c), i32 4, <2 x i1> undef) @@ -50,7 +49,6 @@ define void @masked_scatter5() { define void @masked_scatter6() { ; CHECK-LABEL: define void @masked_scatter6() { -; CHECK-NEXT: store i32 0, ptr @c, align 4 ; CHECK-NEXT: ret void ; call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> zeroinitializer, <2 x ptr> splat (ptr @c), i32 4, <2 x i1> <i1 undef, i1 false>) diff --git a/llvm/test/Transforms/LoopVectorize/single-early-exit-deref-assumptions.ll b/llvm/test/Transforms/LoopVectorize/single-early-exit-deref-assumptions.ll index 9620697..f794620 100644 --- a/llvm/test/Transforms/LoopVectorize/single-early-exit-deref-assumptions.ll +++ b/llvm/test/Transforms/LoopVectorize/single-early-exit-deref-assumptions.ll @@ -512,8 +512,8 @@ define i64 @early_exit_alignment_and_deref_known_via_assumption_with_constant_si ; CHECK-NEXT: [[INDEX1:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[LOOP_INC:.*]] ], [ 0, %[[ENTRY]] ] ; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX1]] ; CHECK-NEXT: [[LD1:%.*]] = load i8, ptr [[ARRAYIDX2]], align 1 -; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX1]] -; CHECK-NEXT: [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX1]] +; CHECK-NEXT: [[LD2:%.*]] = load i8, ptr [[TMP1]], align 1 ; CHECK-NEXT: [[CMP3:%.*]] = icmp eq i8 [[LD1]], [[LD2]] ; CHECK-NEXT: br i1 [[CMP3]], label %[[LOOP_INC]], label %[[LOOP_END:.*]] ; CHECK: [[LOOP_INC]]: diff --git a/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll b/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll index ed0bd3f..cf62fd5 100644 --- a/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll +++ b/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll @@ -55,6 +55,54 @@ entry: ret void } +define void @test_add_udiv(ptr %arr1, ptr %arr2, i32 %a0, i32 %a1, i32 %a2, i32 %a3) { +; CHECK-LABEL: @test_add_udiv( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[GEP1_2:%.*]] = getelementptr i32, ptr [[ARR1:%.*]], i32 2 +; CHECK-NEXT: [[GEP1_3:%.*]] = getelementptr i32, ptr [[ARR1]], i32 3 +; CHECK-NEXT: [[V2:%.*]] = load i32, ptr [[GEP1_2]], align 4 +; CHECK-NEXT: [[V3:%.*]] = load i32, ptr [[GEP1_3]], align 4 +; CHECK-NEXT: [[Y2:%.*]] = add nsw i32 [[A2:%.*]], 42 +; CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[ARR1]], align 4 +; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> <i32 poison, i32 poison, i32 0, i32 poison>, i32 [[A0:%.*]], i32 0 +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> [[TMP1]], i32 [[A1:%.*]], i32 1 +; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x i32> [[TMP2]], i32 [[A3:%.*]], i32 3 +; CHECK-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> <i32 1146, i32 146, i32 0, i32 0>, [[TMP3]] +; CHECK-NEXT: [[RES2:%.*]] = udiv i32 [[V2]], [[Y2]] +; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x i32> poison, i32 [[RES2]], i32 2 +; CHECK-NEXT: [[TMP6:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[V3]], i32 3 +; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <2 x i32> [[TMP0]], <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison> +; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <4 x i32> [[TMP6]], <4 x i32> [[TMP7]], <4 x i32> <i32 4, i32 5, i32 2, i32 3> +; CHECK-NEXT: [[TMP9:%.*]] = add nsw <4 x i32> [[TMP8]], [[TMP4]] +; CHECK-NEXT: store <4 x i32> [[TMP9]], ptr [[ARR2:%.*]], align 4 +; CHECK-NEXT: ret void +; +entry: + %gep1.1 = getelementptr i32, ptr %arr1, i32 1 + %gep1.2 = getelementptr i32, ptr %arr1, i32 2 + %gep1.3 = getelementptr i32, ptr %arr1, i32 3 + %gep2.1 = getelementptr i32, ptr %arr2, i32 1 + %gep2.2 = getelementptr i32, ptr %arr2, i32 2 + %gep2.3 = getelementptr i32, ptr %arr2, i32 3 + %v0 = load i32, ptr %arr1 + %v1 = load i32, ptr %gep1.1 + %v2 = load i32, ptr %gep1.2 + %v3 = load i32, ptr %gep1.3 + %y0 = add nsw i32 %a0, 1146 + %y1 = add nsw i32 %a1, 146 + %y2 = add nsw i32 %a2, 42 + %y3 = add nsw i32 %a3, 0 + %res0 = add nsw i32 %v0, %y0 + %res1 = add nsw i32 %v1, %y1 + %res2 = udiv i32 %v2, %y2 + %res3 = add nsw i32 %v3, %y3 + store i32 %res0, ptr %arr2 + store i32 %res1, ptr %gep2.1 + store i32 %res2, ptr %gep2.2 + store i32 %res3, ptr %gep2.3 + ret void +} + ;; Similar test, but now div/rem is main opcode and not the alternate one. Same issue. define void @test_urem_add(ptr %arr1, ptr %arr2, i32 %a0, i32 %a1, i32 %a2, i32 %a3) { ; CHECK-LABEL: @test_urem_add( @@ -114,3 +162,56 @@ entry: store i32 %res3, ptr %gep2.3 ret void } + +define void @test_srem_add(ptr %arr1, ptr %arr2, i32 %a0, i32 %a1, i32 %a2, i32 %a3) { +; CHECK-LABEL: @test_srem_add( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[GEP1_1:%.*]] = getelementptr i32, ptr [[ARR1:%.*]], i32 1 +; CHECK-NEXT: [[GEP1_2:%.*]] = getelementptr i32, ptr [[ARR1]], i32 2 +; CHECK-NEXT: [[GEP1_3:%.*]] = getelementptr i32, ptr [[ARR1]], i32 3 +; CHECK-NEXT: [[GEP2_1:%.*]] = getelementptr i32, ptr [[ARR2:%.*]], i32 1 +; CHECK-NEXT: [[GEP2_2:%.*]] = getelementptr i32, ptr [[ARR2]], i32 2 +; CHECK-NEXT: [[GEP2_3:%.*]] = getelementptr i32, ptr [[ARR2]], i32 3 +; CHECK-NEXT: [[V0:%.*]] = load i32, ptr [[ARR1]], align 4 +; CHECK-NEXT: [[V1:%.*]] = load i32, ptr [[GEP1_1]], align 4 +; CHECK-NEXT: [[V2:%.*]] = load i32, ptr [[GEP1_2]], align 4 +; CHECK-NEXT: [[V3:%.*]] = load i32, ptr [[GEP1_3]], align 4 +; CHECK-NEXT: [[Y0:%.*]] = add nsw i32 [[A0:%.*]], 1146 +; CHECK-NEXT: [[Y1:%.*]] = add nsw i32 [[A1:%.*]], 146 +; CHECK-NEXT: [[Y2:%.*]] = add nsw i32 [[A2:%.*]], 42 +; CHECK-NEXT: [[Y3:%.*]] = add nsw i32 [[A3:%.*]], 0 +; CHECK-NEXT: [[RES0:%.*]] = srem i32 [[V0]], [[Y0]] +; CHECK-NEXT: [[RES1:%.*]] = srem i32 [[V1]], [[Y1]] +; CHECK-NEXT: [[RES2:%.*]] = srem i32 [[V2]], [[Y2]] +; CHECK-NEXT: [[RES3:%.*]] = add nsw i32 [[V3]], [[Y3]] +; CHECK-NEXT: store i32 [[RES0]], ptr [[ARR2]], align 4 +; CHECK-NEXT: store i32 [[RES1]], ptr [[GEP2_1]], align 4 +; CHECK-NEXT: store i32 [[RES2]], ptr [[GEP2_2]], align 4 +; CHECK-NEXT: store i32 [[RES3]], ptr [[GEP2_3]], align 4 +; CHECK-NEXT: ret void +; +entry: + %gep1.1 = getelementptr i32, ptr %arr1, i32 1 + %gep1.2 = getelementptr i32, ptr %arr1, i32 2 + %gep1.3 = getelementptr i32, ptr %arr1, i32 3 + %gep2.1 = getelementptr i32, ptr %arr2, i32 1 + %gep2.2 = getelementptr i32, ptr %arr2, i32 2 + %gep2.3 = getelementptr i32, ptr %arr2, i32 3 + %v0 = load i32, ptr %arr1 + %v1 = load i32, ptr %gep1.1 + %v2 = load i32, ptr %gep1.2 + %v3 = load i32, ptr %gep1.3 + %y0 = add nsw i32 %a0, 1146 + %y1 = add nsw i32 %a1, 146 + %y2 = add nsw i32 %a2, 42 + %y3 = add nsw i32 %a3, 0 + %res0 = srem i32 %v0, %y0 + %res1 = srem i32 %v1, %y1 + %res2 = srem i32 %v2, %y2 + %res3 = add nsw i32 %v3, %y3 + store i32 %res0, ptr %arr2 + store i32 %res1, ptr %gep2.1 + store i32 %res2, ptr %gep2.2 + store i32 %res3, ptr %gep2.3 + ret void +} diff --git a/llvm/test/tools/llvm-objdump/ELF/AMDGPU/kd-gfx1250.s b/llvm/test/tools/llvm-objdump/ELF/AMDGPU/kd-gfx1250.s index 3e96ea3..13f20bf 100644 --- a/llvm/test/tools/llvm-objdump/ELF/AMDGPU/kd-gfx1250.s +++ b/llvm/test/tools/llvm-objdump/ELF/AMDGPU/kd-gfx1250.s @@ -20,7 +20,7 @@ ; CHECK-NEXT: ; IMAGE_OP 0 ; CHECK-NEXT: .amdhsa_next_free_vgpr 32 ; CHECK-NEXT: .amdhsa_reserve_vcc 0 -; CHECK-NEXT: .amdhsa_reserve_xnack_mask 0 +; CHECK-NEXT: .amdhsa_reserve_xnack_mask 1 ; CHECK-NEXT: .amdhsa_next_free_sgpr 8 ; CHECK-NEXT: .amdhsa_float_round_mode_32 0 ; CHECK-NEXT: .amdhsa_float_round_mode_16_64 0 @@ -76,7 +76,7 @@ ; CHECK-NEXT: ; IMAGE_OP 0 ; CHECK-NEXT: .amdhsa_next_free_vgpr 32 ; CHECK-NEXT: .amdhsa_reserve_vcc 0 -; CHECK-NEXT: .amdhsa_reserve_xnack_mask 0 +; CHECK-NEXT: .amdhsa_reserve_xnack_mask 1 ; CHECK-NEXT: .amdhsa_next_free_sgpr 8 ; CHECK-NEXT: .amdhsa_float_round_mode_32 0 ; CHECK-NEXT: .amdhsa_float_round_mode_16_64 0 |