derek/TheRock
1
0 Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity
Files
1a182e2dac58879f215c0fad0f5f78c76dcafe56
TheRock/compiler/amd-llvm/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w32.cl
T
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
Details
CodeQL / Analyze (python) (push) Canceled after 0s
Details
pre-commit / pre-commit (push) Canceled after 0s
Details
Multi-Arch CI / setup (push) Canceled after 0s
Details
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Details
Setup (Multi-Arch) / setup (push) Canceled after 0s
Details
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Details
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
Details
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Details
Multi-Arch CI / CI Summary (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Details
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Details
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Details
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

174 lines
9.6 KiB
Common Lisp
Raw Blame History

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6
// REQUIRES: amdgpu-registered-target
// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1170 -target-feature +wavefrontsize32 -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1200 -target-feature +wavefrontsize32 -emit-llvm -o - %s | FileCheck %s
typedef int v2i __attribute__((ext_vector_type(2)));
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)));
// Wave32
//
// amdgcn_wmma_f32_16x16x16_f16
//
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_f32_16x16x16_f16_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v8f32.v8f16(<8 x half> [[A]], <8 x half> [[B]], <8 x float> [[C]])
// CHECK-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8:![0-9]+]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_f32_16x16x16_f16_w32(global v8f* out, v8h a, v8h b, v8f c)
{
*out = __builtin_amdgcn_wmma_f32_16x16x16_f16_w32_gfx12(a, b, c);
}
//
// amdgcn_wmma_f32_16x16x16_bf16
//
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_f32_16x16x16_bf16_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <8 x i16> noundef [[A:%.*]], <8 x i16> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf16.v8f32.v8i16(<8 x i16> [[A]], <8 x i16> [[B]], <8 x float> [[C]])
// CHECK-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_f32_16x16x16_bf16_w32(global v8f* out, v8s a, v8s b, v8f c)
{
*out = __builtin_amdgcn_wmma_f32_16x16x16_bf16_w32_gfx12(a, b, c);
}
//
// amdgcn_wmma_f16_16x16x16_f16
//
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_f16_16x16x16_f16_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]], <8 x half> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.v8f16.v8f16(<8 x half> [[A]], <8 x half> [[B]], <8 x half> [[C]], i1 false)
// CHECK-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_f16_16x16x16_f16_w32(global v8h* out, v8h a, v8h b, v8h c)
{
*out = __builtin_amdgcn_wmma_f16_16x16x16_f16_w32_gfx12(a, b, c);
}
//
// amdgcn_wmma_bf16_16x16x16_bf16
//
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_bf16_16x16x16_bf16_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <8 x i16> noundef [[A:%.*]], <8 x i16> noundef [[B:%.*]], <8 x i16> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.v8i16.v8i16(<8 x i16> [[A]], <8 x i16> [[B]], <8 x i16> [[C]], i1 false)
// CHECK-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_bf16_16x16x16_bf16_w32(global v8s* out, v8s a, v8s b, v8s c)
{
*out = __builtin_amdgcn_wmma_bf16_16x16x16_bf16_w32_gfx12(a, b, c);
}
//
// amdgcn_wmma_i32_16x16x16_iu8
//
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_i32_16x16x16_iu8_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu8.v8i32.v2i32(i1 true, <2 x i32> [[A]], i1 true, <2 x i32> [[B]], <8 x i32> [[C]], i1 false)
// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_i32_16x16x16_iu8_w32(global v8i* out, v2i a, v2i b, v8i c)
{
*out = __builtin_amdgcn_wmma_i32_16x16x16_iu8_w32_gfx12(true, a, true, b, c, false);
}
//
// amdgcn_wmma_i32_16x16x16_iu4
//
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_i32_16x16x16_iu4_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu4.v8i32.i32(i1 true, i32 [[A]], i1 true, i32 [[B]], <8 x i32> [[C]], i1 false)
// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_i32_16x16x16_iu4_w32(global v8i* out, int a, int b, v8i c)
{
*out = __builtin_amdgcn_wmma_i32_16x16x16_iu4_w32_gfx12(true, a, true, b, c, false);
}
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_f32_16x16x16_fp8_fp8_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.fp8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]])
// CHECK-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_f32_16x16x16_fp8_fp8_w32(global v8f* out, v2i a, v2i b, v8f c)
{
*out = __builtin_amdgcn_wmma_f32_16x16x16_fp8_fp8_w32_gfx12(a, b, c);
}
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_f32_16x16x16_fp8_bf8_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.bf8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]])
// CHECK-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_f32_16x16x16_fp8_bf8_w32(global v8f* out, v2i a, v2i b, v8f c)
{
*out = __builtin_amdgcn_wmma_f32_16x16x16_fp8_bf8_w32_gfx12(a, b, c);
}
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_f32_16x16x16_bf8_fp8_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.fp8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]])
// CHECK-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_f32_16x16x16_bf8_fp8_w32(global v8f* out, v2i a, v2i b, v8f c)
{
*out = __builtin_amdgcn_wmma_f32_16x16x16_bf8_fp8_w32_gfx12(a, b, c);
}
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_f32_16x16x16_bf8_bf8_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.bf8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]])
// CHECK-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_f32_16x16x16_bf8_bf8_w32(global v8f* out, v2i a, v2i b, v8f c)
{
*out = __builtin_amdgcn_wmma_f32_16x16x16_bf8_bf8_w32_gfx12(a, b, c);
}
// CHECK-LABEL: define dso_local void @test_amdgcn_wmma_i32_16x16x32_iu4_w32(
// CHECK-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x32.iu4.v8i32.v2i32(i1 true, <2 x i32> [[A]], i1 true, <2 x i32> [[B]], <8 x i32> [[C]], i1 false)
// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]]
// CHECK-NEXT: ret void
//
void test_amdgcn_wmma_i32_16x16x32_iu4_w32(global v8i* out, v2i a, v2i b, v8i c)
{
*out = __builtin_amdgcn_wmma_i32_16x16x32_iu4_w32_gfx12(true, a, true, b, c, false);
}
//.
// CHECK: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0}
// CHECK: [[META7]] = !{!"Simple C/C++ TBAA"}
// CHECK: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0}
//.
Reference in new issue View Git Blame Copy Permalink