// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// REQUIRES: amdgpu-registered-target
// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1250 -target-feature +wavefrontsize32 -emit-llvm -o - %s | FileCheck %s --check-prefix=CHECK-GFX1250

typedef float  v16f  __attribute__((ext_vector_type(16)));
typedef float  v8f   __attribute__((ext_vector_type(8)));
typedef float  v2f   __attribute__((ext_vector_type(2)));
typedef half   v8h   __attribute__((ext_vector_type(8)));
typedef half   v16h  __attribute__((ext_vector_type(16)));
typedef half   v32h  __attribute__((ext_vector_type(32)));
typedef __bf16 v32bf16 __attribute__((ext_vector_type(32)));
typedef __bf16 v16bf16 __attribute__((ext_vector_type(16)));
typedef __bf16 v8bf16 __attribute__((ext_vector_type(8)));
typedef int    v16i   __attribute__((ext_vector_type(16)));
typedef int    v8i   __attribute__((ext_vector_type(8)));
typedef int    v2i   __attribute__((ext_vector_type(2)));

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x4_f32(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x4.f32.v8f32.v2f32(<2 x float> [[A:%.*]], <2 x float> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7:![0-9]+]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x4_f32(global v8f* out, v2f a, v2f b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x4_f32(0, a, 0, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x32_bf16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x32.bf16.v8f32.v16bf16(<16 x bfloat> [[A:%.*]], <16 x bfloat> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 true, i1 false)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x32_bf16(global v8f* out, v16bf16 a, v16bf16 b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x32_bf16(0, a, 0, b, 0, c, true, false);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_bf16_16x16x32_bf16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x bfloat> @llvm.amdgcn.wmma.bf16.16x16x32.bf16.v8bf16.v16bf16(<16 x bfloat> [[A:%.*]], <16 x bfloat> [[B:%.*]], i16 0, <8 x bfloat> [[C:%.*]], i1 false, i1 false)
// CHECK-GFX1250-NEXT:    store <8 x bfloat> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_bf16_16x16x32_bf16(global v8bf16* out, v16bf16 a, v16bf16 b, v8bf16 c)
{
  *out = __builtin_amdgcn_wmma_bf16_16x16x32_bf16(0, a, 0, b, 0, c, false, false);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_bf16f32_16x16x32_bf16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x bfloat> @llvm.amdgcn.wmma.bf16f32.16x16x32.bf16.v8bf16.v16bf16.v8f32(<16 x bfloat> [[A:%.*]], <16 x bfloat> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x bfloat> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_bf16f32_16x16x32_bf16(global v8bf16* out, v16bf16 a, v16bf16 b, v8f c)
{
  *out = __builtin_amdgcn_wmma_bf16f32_16x16x32_bf16(0, a, 0, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x64_fp8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x64.fp8.fp8.v8f32.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 true, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x64_fp8_fp8(global v8f* out, v8i a, v8i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x64_fp8_fp8(a, b, 0, c, true, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x64_fp8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x64.fp8.bf8.v8f32.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x64_fp8_bf8(global v8f* out, v8i a, v8i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x64_fp8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x64_bf8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x64.bf8.fp8.v8f32.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x64_bf8_fp8(global v8f* out, v8i a, v8i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x64_bf8_fp8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x64_bf8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x64.bf8.bf8.v8f32.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x64_bf8_bf8(global v8f* out, v8i a, v8i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x64_bf8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x64_fp8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x64.fp8.fp8.v8f16.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x64_fp8_fp8(global v8h* out, v8i a, v8i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x64_fp8_fp8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x64_fp8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x64.fp8.bf8.v8f16.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x64_fp8_bf8(global v8h* out, v8i a, v8i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x64_fp8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x64_bf8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x64.bf8.fp8.v8f16.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x64_bf8_fp8(global v8h* out, v8i a, v8i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x64_bf8_fp8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x64_bf8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x64.bf8.bf8.v8f16.v8i32(<8 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x64_bf8_bf8(global v8h* out, v8i a, v8i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x64_bf8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_i32_16x16x64_iu8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x64.iu8.v8i32.v8i32(i1 false, <8 x i32> [[A:%.*]], i1 false, <8 x i32> [[B:%.*]], <8 x i32> [[C:%.*]], i1 false, i1 true, i1 false)
// CHECK-GFX1250-NEXT:    store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_i32_16x16x64_iu8(global v8i* out, v8i a, v8i b, v8i c)
{
  *out = __builtin_amdgcn_wmma_i32_16x16x64_iu8(0, a, 0, b, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_i32_16x16x64_iu8_clamp(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x64.iu8.v8i32.v8i32(i1 false, <8 x i32> [[A:%.*]], i1 false, <8 x i32> [[B:%.*]], <8 x i32> [[C:%.*]], i1 false, i1 true, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_i32_16x16x64_iu8_clamp(global v8i* out, v8i a, v8i b, v8i c)
{
  *out = __builtin_amdgcn_wmma_i32_16x16x64_iu8(0, a, 0, b, c, false, true, 1);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x128_f8f6f4(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = shufflevector <16 x i32> [[B:%.*]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
// CHECK-GFX1250-NEXT:    [[TMP1:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x128.f8f6f4.v8f32.v16i32.v12i32(i32 1, <16 x i32> [[A:%.*]], i32 2, <12 x i32> [[TMP0]], i16 0, <8 x float> [[C:%.*]])
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP1]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = shufflevector <16 x i32> [[B:%.*]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
// CHECK-GFX1250-NEXT:    [[TMP1:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.scale.f32.16x16x128.f8f6f4.v8f32.v16i32.v12i32(i32 1, <16 x i32> [[A:%.*]], i32 2, <12 x i32> [[TMP0]], i16 0, <8 x float> [[C:%.*]], i32 1, i32 2, i32 [[SCALE_SRC0:%.*]], i32 2, i32 1, i32 [[SCALE_SRC1:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP1]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f c, int scale_src0, int scale_src1)
{
  *out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 2, scale_src0, 2, 1, scale_src1, 0, 1);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = shufflevector <16 x i32> [[B:%.*]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
// CHECK-GFX1250-NEXT:    [[TMP1:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.scale16.f32.16x16x128.f8f6f4.v8f32.v16i32.v12i32(i32 1, <16 x i32> [[A:%.*]], i32 2, <12 x i32> [[TMP0]], i16 0, <8 x float> [[C:%.*]], i32 1, i32 2, i64 [[SCALE_SRC0:%.*]], i32 2, i32 1, i64 [[SCALE_SRC1:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP1]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f c, long scale_src0, long scale_src1)
{
  *out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 2, scale_src0, 2, 1, scale_src1, 0, 1);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x32_f16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x32.f16.v8f32.v16f16(<16 x half> [[A:%.*]], <16 x half> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x32_f16(global v8f* out, v16h a, v16h b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x32_f16(0, a, 0, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x32_f16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x32.f16.v8f16.v16f16(<16 x half> [[A:%.*]], <16 x half> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x32_f16(global v8h* out, v16h a, v16h b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x32_f16(0, a, 0, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x128_fp8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x128.fp8.fp8.v8f16.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x128_fp8_fp8(global v8h* out, v16i a, v16i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x128_fp8_fp8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x128_fp8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x128.fp8.bf8.v8f16.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x128_fp8_bf8(global v8h* out, v16i a, v16i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x128_fp8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x128_bf8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x128.bf8.fp8.v8f16.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x128_bf8_fp8(global v8h* out, v16i a, v16i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x128_bf8_fp8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f16_16x16x128_bf8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x128.bf8.bf8.v8f16.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x half> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f16_16x16x128_bf8_bf8(global v8h* out, v16i a, v16i b, v8h c)
{
  *out = __builtin_amdgcn_wmma_f16_16x16x128_bf8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x128_fp8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x128.fp8.fp8.v8f32.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 true, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x128_fp8_fp8(global v8f* out, v16i a, v16i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x128_fp8_fp8(a, b, 0, c, true, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x128_fp8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x128.fp8.bf8.v8f32.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x128_fp8_bf8(global v8f* out, v16i a, v16i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x128_fp8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x128_bf8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x128.bf8.fp8.v8f32.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x128_bf8_fp8(global v8f* out, v16i a, v16i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x128_bf8_fp8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x128_bf8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x128.bf8.bf8.v8f32.v16i32(<16 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_16x16x128_bf8_bf8(global v8f* out, v16i a, v16i b, v8f c)
{
  *out = __builtin_amdgcn_wmma_f32_16x16x128_bf8_bf8(a, b, 0, c, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_wmma_f32_32x16x128_f4(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <16 x float> @llvm.amdgcn.wmma.f32.32x16x128.f4.v16f32.v16i32.v8i32(<16 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <16 x float> [[C:%.*]])
// CHECK-GFX1250-NEXT:    store <16 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 64, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_f32_wmma_f32_32x16x128_f4(global v16f* out, v16i a, v8i b, v16f c)
{
  *out = __builtin_amdgcn_wmma_f32_32x16x128_f4(a, b, 0, c);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_scale_f32_32x16x128_f4(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <16 x float> @llvm.amdgcn.wmma.scale.f32.32x16x128.f4.v16f32.v16i32.v8i32(<16 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <16 x float> [[C:%.*]], i32 1, i32 2, i32 [[SCALE_SRC0:%.*]], i32 2, i32 1, i32 [[SCALE_SRC1:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <16 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 64, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_scale_f32_32x16x128_f4(global v16f* out, v16i a, v8i b, v16f c, int scale_src0, int scale_src1)
{
  *out = __builtin_amdgcn_wmma_scale_f32_32x16x128_f4(a, b, 0, c, 1, 2, scale_src0, 2, 1, scale_src1, 0, 1);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_scale16_f32_32x16x128_f4(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <16 x float> @llvm.amdgcn.wmma.scale16.f32.32x16x128.f4.v16f32.v16i32.v8i32(<16 x i32> [[A:%.*]], <8 x i32> [[B:%.*]], i16 0, <16 x float> [[C:%.*]], i32 1, i32 2, i64 [[SCALE_SRC0:%.*]], i32 2, i32 1, i64 [[SCALE_SRC1:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <16 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 64, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_wmma_scale16_f32_32x16x128_f4(global v16f* out, v16i a, v8i b, v16f c, long scale_src0, long scale_src1)
{
  *out = __builtin_amdgcn_wmma_scale16_f32_32x16x128_f4(a, b, 0, c, 1, 2, scale_src0, 2, 1, scale_src1, 0, 1);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f32_16x16x64_bf16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x64.bf16.v8f32.v16bf16.v32bf16.i32(i1 false, <16 x bfloat> [[A:%.*]], i1 false, <32 x bfloat> [[B:%.*]], <8 x float> [[C:%.*]], i32 [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x64_bf16(global v8f* out, v16bf16 a, v32bf16 b, v8f c, int index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x64_bf16(0, a, 0, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_bf16_16x16x64_bf16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x bfloat> @llvm.amdgcn.swmmac.bf16.16x16x64.bf16.v8bf16.v16bf16.v32bf16.i32(i1 false, <16 x bfloat> [[A:%.*]], i1 false, <32 x bfloat> [[B:%.*]], <8 x bfloat> [[C:%.*]], i32 [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x bfloat> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_bf16_16x16x64_bf16(global v8bf16* out, v16bf16 a, v32bf16 b, v8bf16 c, int index)
{
  *out = __builtin_amdgcn_swmmac_bf16_16x16x64_bf16(0, a, 0, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_bf16f32_16x16x64_bf16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.bf16f32.16x16x64.bf16.v8f32.v16bf16.v32bf16.i32(i1 false, <16 x bfloat> [[A:%.*]], i1 false, <32 x bfloat> [[B:%.*]], <8 x float> [[C:%.*]], i32 [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_bf16f32_16x16x64_bf16(global v8f* out, v16bf16 a, v32bf16 b, v8f c, int index)
{
  *out = __builtin_amdgcn_swmmac_bf16f32_16x16x64_bf16(0, a, 0, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f32_16x16x128_fp8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x128.fp8.fp8.v8f32.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x float> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x128_fp8_fp8(global v8f* out, v8i a, v16i b, v8f c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x128_fp8_fp8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f32_16x16x128_fp8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x128.fp8.bf8.v8f32.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x float> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x128_fp8_bf8(global v8f* out, v8i a, v16i b, v8f c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x128_fp8_bf8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f32_16x16x128_bf8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x128.bf8.fp8.v8f32.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x float> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x128_bf8_fp8(global v8f* out, v8i a, v16i b, v8f c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x128_bf8_fp8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f32_16x16x128_bf8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x128.bf8.bf8.v8f32.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x float> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x128_bf8_bf8(global v8f* out, v8i a, v16i b, v8f c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x128_bf8_bf8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f16_16x16x128_fp8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.swmmac.f16.16x16x128.fp8.fp8.v8f16.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x half> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f16_16x16x128_fp8_fp8(global v8h* out, v8i a, v16i b, v8h c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f16_16x16x128_fp8_fp8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f16_16x16x128_fp8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.swmmac.f16.16x16x128.fp8.bf8.v8f16.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x half> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f16_16x16x128_fp8_bf8(global v8h* out, v8i a, v16i b, v8h c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f16_16x16x128_fp8_bf8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f16_16x16x128_bf8_fp8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.swmmac.f16.16x16x128.bf8.fp8.v8f16.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x half> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f16_16x16x128_bf8_fp8(global v8h* out, v8i a, v16i b, v8h c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f16_16x16x128_bf8_fp8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f16_16x16x128_bf8_bf8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.swmmac.f16.16x16x128.bf8.bf8.v8f16.v8i32.v16i32.v2i32(<8 x i32> [[A:%.*]], <16 x i32> [[B:%.*]], <8 x half> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f16_16x16x128_bf8_bf8(global v8h* out, v8i a, v16i b, v8h c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_f16_16x16x128_bf8_bf8(a, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_i32_16x16x128_iu8(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x128.iu8.v8i32.v8i32.v16i32.v2i32(i1 false, <8 x i32> [[A:%.*]], i1 false, <16 x i32> [[B:%.*]], <8 x i32> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true, i1 false)
// CHECK-GFX1250-NEXT:    store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_i32_16x16x128_iu8(global v8i* out, v8i a, v16i b, v8i c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_i32_16x16x128_iu8(0, a, 0, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_i32_16x16x128_iu8_clamp(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x128.iu8.v8i32.v8i32.v16i32.v2i32(i1 false, <8 x i32> [[A:%.*]], i1 false, <16 x i32> [[B:%.*]], <8 x i32> [[C:%.*]], <2 x i32> [[INDEX:%.*]], i1 false, i1 true, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_i32_16x16x128_iu8_clamp(global v8i* out, v8i a, v16i b, v8i c, v2i index)
{
  *out = __builtin_amdgcn_swmmac_i32_16x16x128_iu8(0, a, 0, b, c, index, false, true, 1);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f32_16x16x64_f16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x64.f16.v8f32.v16f16.v32f16.i32(i1 false, <16 x half> [[A:%.*]], i1 false, <32 x half> [[B:%.*]], <8 x float> [[C:%.*]], i32 [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x64_f16(global v8f* out, v16h a, v32h b, v8f c, int index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x64_f16(0, a, 0, b, c, index, false, true);
}

// CHECK-GFX1250-LABEL: @test_amdgcn_swmmac_f16_16x16x64_f16(
// CHECK-GFX1250-NEXT:  entry:
// CHECK-GFX1250-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.swmmac.f16.16x16x64.f16.v8f16.v16f16.v32f16.i32(i1 false, <16 x half> [[A:%.*]], i1 false, <32 x half> [[B:%.*]], <8 x half> [[C:%.*]], i32 [[INDEX:%.*]], i1 false, i1 true)
// CHECK-GFX1250-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA7]]
// CHECK-GFX1250-NEXT:    ret void
//
void test_amdgcn_swmmac_f16_16x16x64_f16(global v8h* out, v16h a, v32h b, v8h c, int index)
{
  *out = __builtin_amdgcn_swmmac_f16_16x16x64_f16(0, a, 0, b, c, index, false, true);
}