diff options
Diffstat (limited to 'clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c')
| -rw-r--r-- | clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c | 411 |
1 files changed, 308 insertions, 103 deletions
diff --git a/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c b/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c index 8fc9a68..f59a964 100644 --- a/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c +++ b/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c @@ -1,17 +1,26 @@ -// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6 // Update then manual applied to commonize the checks for AIX and LoP. // RUN: %clang_cc1 -O3 -triple powerpc64le-unknown-unknown -target-cpu future \ // RUN: -emit-llvm %s -o - | FileCheck %s // RUN: %clang_cc1 -O3 -triple powerpc64-ibm-aix -target-cpu future \ -// RUN: -emit-llvm %s -o - | FileCheck %s +// RUN: -emit-llvm %s -o - | FileCheck %s --check-prefix=AIX -// CHECK-LABEL: void @test_dmxvbf16gerx2( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2:![0-9]+]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8:![0-9]+]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8:![0-9]+]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -19,14 +28,24 @@ void test_dmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigne *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -34,14 +53,24 @@ void test_dmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2np( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -49,14 +78,24 @@ void test_dmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -64,14 +103,24 @@ void test_dmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -79,13 +128,22 @@ void test_dmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -93,14 +151,24 @@ void test_pmdmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -108,14 +176,24 @@ void test_pmdmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2np( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -123,14 +201,24 @@ void test_pmdmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -138,14 +226,24 @@ void test_pmdmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -153,13 +251,22 @@ void test_pmdmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2:![0-9]+]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -167,14 +274,24 @@ void test_dmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -182,14 +299,24 @@ void test_dmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2np( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -197,14 +324,24 @@ void test_dmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -212,14 +349,24 @@ void test_dmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -227,13 +374,22 @@ void test_dmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -241,14 +397,24 @@ void test_pmdmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -256,14 +422,24 @@ void test_pmdmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2np( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -271,14 +447,24 @@ void test_pmdmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -286,14 +472,24 @@ void test_pmdmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -301,9 +497,18 @@ void test_pmdmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__vector_pair", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +//. +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK: [[META7]] = !{!"__dmr1024", [[META4]], i64 0} +// CHECK: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// CHECK: [[__DMR1024_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK: [[META9]] = !{!"__dmr1024", [[META4]], i64 0} +//. +// AIX: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// AIX: [[META5]] = !{!"Simple C/C++ TBAA"} +// AIX: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// AIX: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// AIX: [[__DMR1024_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// AIX: [[META9]] = !{!"__dmr1024", [[META4]], i64 0} +//. |