// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6
// 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 --check-prefix=AIX


// CHECK-LABEL: define dso_local void @test_dmxvi8gerx4(
// 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 [[__VECTOR_PAIR_TBAA6:![0-9]+]]
// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvi8gerx4(<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_dmxvi8gerx4(
// 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.dmxvi8gerx4(<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_dmxvi8gerx4(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
  __vector_pair vp = *((__vector_pair *)vpp);
  __builtin_mma_dmxvi8gerx4(&vdmr, vp, vc);
  *((__dmr1024 *)resp) = vdmr;
}

// CHECK-LABEL: define dso_local void @test_pmdmxvi8gerx4(
// 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 [[__VECTOR_PAIR_TBAA6]]
// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvi8gerx4(<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_pmdmxvi8gerx4(
// 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.pmdmxvi8gerx4(<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_pmdmxvi8gerx4(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
  __vector_pair vp = *((__vector_pair *)vpp);
  __builtin_mma_pmdmxvi8gerx4(&vdmr, vp, vc, 0, 0, 0);
  *((__dmr1024 *)resp) = vdmr;
}

// CHECK-LABEL: define dso_local void @test_dmxvi8gerx4pp(
// 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 [[__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.dmxvi8gerx4pp(<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_dmxvi8gerx4pp(
// 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.dmxvi8gerx4pp(<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_dmxvi8gerx4pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
  __vector_pair vp = *((__vector_pair *)vpp);
  __builtin_mma_dmxvi8gerx4pp(&vdmr, vp, vc);
  *((__dmr1024 *)resp) = vdmr;
}

// CHECK-LABEL: define dso_local void @test_pmdmxvi8gerx4pp(
// 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 [[__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.pmdmxvi8gerx4pp(<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_pmdmxvi8gerx4pp(
// 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.pmdmxvi8gerx4pp(<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_pmdmxvi8gerx4pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
  __vector_pair vp = *((__vector_pair *)vpp);
  __builtin_mma_pmdmxvi8gerx4pp(&vdmr, vp, vc, 0, 0, 0);
  *((__dmr1024 *)resp) = vdmr;
}

// CHECK-LABEL: define dso_local void @test_dmxvi8gerx4spp(
// 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 [[__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.dmxvi8gerx4spp(<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_dmxvi8gerx4spp(
// 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.dmxvi8gerx4spp(<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_dmxvi8gerx4spp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
  __vector_pair vp = *((__vector_pair *)vpp);
  __builtin_mma_dmxvi8gerx4spp(&vdmr, vp, vc);
  *((__dmr1024 *)resp) = vdmr;
}

// CHECK-LABEL: define dso_local void @test_pmdmxvi8gerx4spp(
// 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 [[__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.pmdmxvi8gerx4spp(<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_pmdmxvi8gerx4spp(
// 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.pmdmxvi8gerx4spp(<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_pmdmxvi8gerx4spp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
  __vector_pair vp = *((__vector_pair *)vpp);
  __builtin_mma_pmdmxvi8gerx4spp(&vdmr, vp, vc, 0, 0, 0);
  *((__dmr1024 *)resp) = vdmr;
}

// CHECK-LABEL: define dso_local void @test_dmf_basic(
// CHECK-SAME: ptr noundef readonly captures(none) [[P:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES1:%.*]], ptr noundef captures(none) [[RES2:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmsetdmrz()
// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmmr(<1024 x i1> [[TMP0]])
// CHECK-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RES1]], align 128
// CHECK-NEXT:    [[TMP2:%.*]] = load <1024 x i1>, ptr [[RES2]], align 128
// CHECK-NEXT:    [[TMP3:%.*]] = load <1024 x i1>, ptr [[P]], align 128
// CHECK-NEXT:    [[TMP4:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxor(<1024 x i1> [[TMP2]], <1024 x i1> [[TMP3]])
// CHECK-NEXT:    store <1024 x i1> [[TMP4]], ptr [[RES2]], align 128
// CHECK-NEXT:    ret void
//
// AIX-LABEL: define void @test_dmf_basic(
// AIX-SAME: ptr noundef readonly captures(none) [[P:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES1:%.*]], ptr noundef captures(none) [[RES2:%.*]]) local_unnamed_addr #[[ATTR0]] {
// AIX-NEXT:  [[ENTRY:.*:]]
// AIX-NEXT:    [[TMP0:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmsetdmrz()
// AIX-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmmr(<1024 x i1> [[TMP0]])
// AIX-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RES1]], align 128
// AIX-NEXT:    [[TMP2:%.*]] = load <1024 x i1>, ptr [[RES2]], align 128
// AIX-NEXT:    [[TMP3:%.*]] = load <1024 x i1>, ptr [[P]], align 128
// AIX-NEXT:    [[TMP4:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxor(<1024 x i1> [[TMP2]], <1024 x i1> [[TMP3]])
// AIX-NEXT:    store <1024 x i1> [[TMP4]], ptr [[RES2]], align 128
// AIX-NEXT:    ret void
//
void test_dmf_basic(char *p, char *res1, char *res2) {
  __dmr1024 x[2];
  __builtin_mma_dmsetdmrz(&x[0]);
  __builtin_mma_dmmr((__dmr1024*)res1, &x[0]);
  __builtin_mma_dmxor((__dmr1024*)res2, (__dmr1024*)p);
}

// CHECK-LABEL: define dso_local void @test_dmf_basic2(
// CHECK-SAME: ptr noundef readonly captures(none) [[P1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES2:%.*]], ptr noundef readonly captures(none) [[V:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = load <16 x i8>, ptr [[V]], align 16, !tbaa [[CHAR_TBAA10:![0-9]+]]
// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.build.dmr(<16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]])
// CHECK-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RES2]], align 128
// CHECK-NEXT:    [[TMP2:%.*]] = load <1024 x i1>, ptr [[P1]], align 128
// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RES1]], align 128
// CHECK-NEXT:    ret void
//
// AIX-LABEL: define void @test_dmf_basic2(
// AIX-SAME: ptr noundef readonly captures(none) [[P1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES2:%.*]], ptr noundef readonly captures(none) [[V:%.*]]) local_unnamed_addr #[[ATTR0]] {
// AIX-NEXT:  [[ENTRY:.*:]]
// AIX-NEXT:    [[TMP0:%.*]] = load <16 x i8>, ptr [[V]], align 16, !tbaa [[CHAR_TBAA10:![0-9]+]]
// AIX-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.build.dmr(<16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]])
// AIX-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RES2]], align 128
// AIX-NEXT:    [[TMP2:%.*]] = load <1024 x i1>, ptr [[P1]], align 128
// AIX-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RES1]], align 128
// AIX-NEXT:    ret void
//
void test_dmf_basic2(char *p1, char *res1, char *res2,
                     vector unsigned char *v) {
  vector unsigned char vv = *v;
  __builtin_mma_build_dmr((__dmr1024*)res2, vv, vv, vv, vv, vv, vv, vv, vv);
  __builtin_mma_disassemble_dmr(res1, (__dmr1024*)p1);
}

// CHECK-LABEL: define dso_local void @test_dmsha2hash(
// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP1:%.*]], ptr noundef readonly captures(none) [[VDMRP2:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP1]], align 128, !tbaa [[__DMR1024_TBAA8]]
// CHECK-NEXT:    [[TMP1:%.*]] = load <1024 x i1>, ptr [[VDMRP2]], align 128, !tbaa [[__DMR1024_TBAA8]]
// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmsha2hash(<1024 x i1> [[TMP0]], <1024 x i1> [[TMP1]], i32 1)
// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]]
// CHECK-NEXT:    ret void
//
// AIX-LABEL: define void @test_dmsha2hash(
// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP1:%.*]], ptr noundef readonly captures(none) [[VDMRP2:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] {
// AIX-NEXT:  [[ENTRY:.*:]]
// AIX-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP1]], align 128, !tbaa [[__DMR1024_TBAA8]]
// AIX-NEXT:    [[TMP1:%.*]] = load <1024 x i1>, ptr [[VDMRP2]], align 128, !tbaa [[__DMR1024_TBAA8]]
// AIX-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmsha2hash(<1024 x i1> [[TMP0]], <1024 x i1> [[TMP1]], i32 1)
// AIX-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]]
// AIX-NEXT:    ret void
//
void test_dmsha2hash(unsigned char *vdmrp1, unsigned char *vdmrp2, unsigned char *resp) {
  __dmr1024 vdmr1 = *((__dmr1024 *)vdmrp1);
  __dmr1024 vdmr2 = *((__dmr1024 *)vdmrp2);
  __builtin_mma_dmsha2hash(&vdmr1, &vdmr2, 1);
  *((__dmr1024 *)resp) = vdmr1;
}

// CHECK-LABEL: define dso_local void @test_dmsha3hash(
// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRPP:%.*]], ptr noundef writeonly captures(none) initializes((0, 256)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT:  [[ENTRY:.*:]]
// CHECK-NEXT:    [[TMP0:%.*]] = load <2048 x i1>, ptr [[VDMRPP]], align 256, !tbaa [[__DMR2048_TBAA11:![0-9]+]]
// CHECK-NEXT:    [[TMP1:%.*]] = tail call <2048 x i1> @llvm.ppc.mma.dmsha3hash(<2048 x i1> [[TMP0]], i32 4)
// CHECK-NEXT:    store <2048 x i1> [[TMP1]], ptr [[RESP]], align 256, !tbaa [[__DMR2048_TBAA11]]
// CHECK-NEXT:    ret void
//
// AIX-LABEL: define void @test_dmsha3hash(
// AIX-SAME: ptr noundef readonly captures(none) [[VDMRPP:%.*]], ptr noundef writeonly captures(none) initializes((0, 256)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] {
// AIX-NEXT:  [[ENTRY:.*:]]
// AIX-NEXT:    [[TMP0:%.*]] = load <2048 x i1>, ptr [[VDMRPP]], align 256, !tbaa [[__DMR2048_TBAA11:![0-9]+]]
// AIX-NEXT:    [[TMP1:%.*]] = tail call <2048 x i1> @llvm.ppc.mma.dmsha3hash(<2048 x i1> [[TMP0]], i32 4)
// AIX-NEXT:    store <2048 x i1> [[TMP1]], ptr [[RESP]], align 256, !tbaa [[__DMR2048_TBAA11]]
// AIX-NEXT:    ret void
//
void test_dmsha3hash(unsigned char *vdmrpp,  unsigned char *resp) {
  __dmr2048 vdmrp = *((__dmr2048 *)vdmrpp);
  __builtin_mma_dmsha3hash(&vdmrp, 4);
  *((__dmr2048 *)resp) = vdmrp;
}

// CHECK-LABEL: define dso_local void @test_dmxxshapad(
// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], <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 [[__DMR1024_TBAA8]]
// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxxshapad(<1024 x i1> [[TMP0]], <16 x i8> [[VC]], i32 2, i32 1, i32 5)
// CHECK-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]]
// CHECK-NEXT:    ret void
//
// AIX-LABEL: define void @test_dmxxshapad(
// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], <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:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxxshapad(<1024 x i1> [[TMP0]], <16 x i8> [[VC]], i32 2, i32 1, i32 5)
// AIX-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]]
// AIX-NEXT:    ret void
//
void test_dmxxshapad(unsigned char *vdmrp, vector unsigned char vc, unsigned char *resp) {
  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
  __builtin_mma_dmxxshapad(&vdmr, vc, 2, 1, 5);
  *((__dmr1024 *)resp) = vdmr;
}
//.
// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0}
// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"}
// 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}
// CHECK: [[CHAR_TBAA10]] = !{[[META4]], [[META4]], i64 0}
// CHECK: [[__DMR2048_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0}
// CHECK: [[META12]] = !{!"__dmr2048", [[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}
// AIX: [[CHAR_TBAA10]] = !{[[META4]], [[META4]], i64 0}
// AIX: [[__DMR2048_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0}
// AIX: [[META12]] = !{!"__dmr2048", [[META4]], i64 0}
//.