// 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 gfx1200 -target-feature +wavefrontsize32 -S -emit-llvm -o - %s | FileCheck %s --check-prefix=CHECK-GFX1200

typedef int    v2i   __attribute__((ext_vector_type(2)));
typedef int    v4i   __attribute__((ext_vector_type(4)));
typedef float  v8f   __attribute__((ext_vector_type(8)));
typedef half   v8h   __attribute__((ext_vector_type(8)));
typedef short  v8s   __attribute__((ext_vector_type(8)));
typedef int    v8i   __attribute__((ext_vector_type(8)));
typedef half  v16h   __attribute__((ext_vector_type(16)));
typedef short v16s   __attribute__((ext_vector_type(16)));

// Wave32

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_f32_16x16x32_f16_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.f16.v8f32.v8f16.v16f16.i16(<8 x half> [[A:%.*]], <16 x half> [[B:%.*]], <8 x float> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4:![0-9]+]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x32_f16_w32(global v8f* out, v8h a, v16h b, v8f c, short index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x32_f16_w32(a, b, c, index);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_f32_16x16x32_bf16_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf16.v8f32.v8i16.v16i16.i16(<8 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <8 x float> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x32_bf16_w32(global v8f* out, v8s a, v16s b, v8f c, short index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x32_bf16_w32(a, b, c, index);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_f16_16x16x32_f16_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.swmmac.f16.16x16x32.f16.v8f16.v8f16.v16f16.i16(<8 x half> [[A:%.*]], <16 x half> [[B:%.*]], <8 x half> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_f16_16x16x32_f16_w32(global v8h* out, v8h a, v16h b, v8h c, short index)
{
  *out = __builtin_amdgcn_swmmac_f16_16x16x32_f16_w32(a, b, c, index);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_bf16_16x16x32_bf16_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x i16> @llvm.amdgcn.swmmac.bf16.16x16x32.bf16.v8i16.v8i16.v16i16.i16(<8 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <8 x i16> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 16, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_bf16_16x16x32_bf16_w32(global v8s* out, v8s a, v16s b, v8s c, short index)
{
  *out = __builtin_amdgcn_swmmac_bf16_16x16x32_bf16_w32(a, b, c, index);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_i32_16x16x32_iu8_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu8.v8i32.v2i32.v4i32.i16(i1 true, <2 x i32> [[A:%.*]], i1 true, <4 x i32> [[B:%.*]], <8 x i32> [[C:%.*]], i16 [[INDEX:%.*]], i1 true)
// CHECK-GFX1200-NEXT:    store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_i32_16x16x32_iu8_w32(global v8i* out, v2i a, v4i b, v8i c, short index)
{
  *out = __builtin_amdgcn_swmmac_i32_16x16x32_iu8_w32(true, a, true, b, c, index, true);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_i32_16x16x32_iu4_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu4.v8i32.i32.v2i32.i16(i1 true, i32 [[A:%.*]], i1 true, <2 x i32> [[B:%.*]], <8 x i32> [[C:%.*]], i16 [[INDEX:%.*]], i1 true)
// CHECK-GFX1200-NEXT:    store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_i32_16x16x32_iu4_w32(global v8i* out, int a, v2i b, v8i c, short index)
{
  *out = __builtin_amdgcn_swmmac_i32_16x16x32_iu4_w32(true, a, true, b, c, index, true);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_i32_16x16x64_iu4_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x64.iu4.v8i32.v2i32.v4i32.i16(i1 true, <2 x i32> [[A:%.*]], i1 true, <4 x i32> [[B:%.*]], <8 x i32> [[C:%.*]], i16 [[INDEX:%.*]], i1 true)
// CHECK-GFX1200-NEXT:    store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_i32_16x16x64_iu4_w32(global v8i* out, v2i a, v4i b, v8i c, short index)
{
  *out = __builtin_amdgcn_swmmac_i32_16x16x64_iu4_w32(true, a, true, b, c, index, true);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_f32_16x16x32_fp8_fp8_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.fp8.v8f32.v2i32.v4i32.i16(<2 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <8 x float> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x32_fp8_fp8_w32(global v8f* out, v2i a, v4i b, v8f c, short index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x32_fp8_fp8_w32(a, b, c, index);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.bf8.v8f32.v2i32.v4i32.i16(<2 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <8 x float> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w32(global v8f* out, v2i a, v4i b, v8f c, short index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w32(a, b, c, index);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.fp8.v8f32.v2i32.v4i32.i16(<2 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <8 x float> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w32(global v8f* out, v2i a, v4i b, v8f c, short index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w32(a, b, c, index);
}

// CHECK-GFX1200-LABEL: @test_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w32(
// CHECK-GFX1200-NEXT:  entry:
// CHECK-GFX1200-NEXT:    [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.bf8.v8f32.v2i32.v4i32.i16(<2 x i32> [[A:%.*]], <4 x i32> [[B:%.*]], <8 x float> [[C:%.*]], i16 [[INDEX:%.*]])
// CHECK-GFX1200-NEXT:    store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
// CHECK-GFX1200-NEXT:    ret void
//
void test_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w32(global v8f* out, v2i a, v4i b, v8f c, short index)
{
  *out = __builtin_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w32(a, b, c, index);
}