derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
80 lines
4.6 KiB
LLVM
80 lines
4.6 KiB
LLVM
; RUN: llvm-as < %s | llvm-dis | FileCheck %s
|
|
|
|
define amdgpu_ps void @test_wmma_f32_16x16x32_bf16(<16 x bfloat> %A, <16 x bfloat> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-LABEL: define amdgpu_ps void @test_wmma_f32_16x16x32_bf16(<16 x bfloat> %A, <16 x bfloat> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: %res = 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 false, i1 true)
|
|
; CHECK-NEXT: store <8 x float> %res, ptr addrspace(1) %out, align 32
|
|
; CHECK-NEXT: ret void
|
|
; CHECK-NEXT: }
|
|
bb:
|
|
%res = call <8 x float> @llvm.amdgcn.wmma.f32.16x16x32.bf16.v8f32.v16bf16(i1 0, <16 x bfloat> %A, i1 0, <16 x bfloat> %B, i16 0, <8 x float> %C, i1 false, i1 true)
|
|
store <8 x float> %res, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_ps void @test_wmma_f32_16x16x4_f32(<2 x float> %A, <2 x float> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-LABEL: define amdgpu_ps void @test_wmma_f32_16x16x4_f32(<2 x float> %A, <2 x float> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: %res = 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-NEXT: store <8 x float> %res, ptr addrspace(1) %out
|
|
; CHECK-NEXT: ret void
|
|
; CHECK-NEXT: }
|
|
bb:
|
|
%res = call <8 x float> @llvm.amdgcn.wmma.f32.16x16x4.f32.v8f32.v2f32(i1 0, <2 x float> %A, i1 0, <2 x float> %B, i16 0, <8 x float> %C, i1 false, i1 true)
|
|
store <8 x float> %res, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_ps void @test_wmma_bf16_16x16x32_bf16(<16 x bfloat> %A, <16 x bfloat> %B, <8 x bfloat> %C, ptr addrspace(1) %out) {
|
|
; CHECK-LABEL: define amdgpu_ps void @test_wmma_bf16_16x16x32_bf16(<16 x bfloat> %A, <16 x bfloat> %B, <8 x bfloat> %C, ptr addrspace(1) %out) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: %res = 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 true)
|
|
; CHECK-NEXT: store <8 x bfloat> %res, ptr addrspace(1) %out
|
|
; CHECK-NEXT: ret void
|
|
; CHECK-NEXT: }
|
|
bb:
|
|
%res = call <8 x bfloat> @llvm.amdgcn.wmma.bf16.16x16x32.bf16.v8bf16.v16bf16(i1 0, <16 x bfloat> %A, i1 0, <16 x bfloat> %B, i16 0, <8 x bfloat> %C, i1 false, i1 true)
|
|
store <8 x bfloat> %res, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_ps void @test_wmma_f32_16x16x32_f16(<16 x half> %A, <16 x half> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-LABEL: define amdgpu_ps void @test_wmma_f32_16x16x32_f16(<16 x half> %A, <16 x half> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: %res = 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-NEXT: store <8 x float> %res, ptr addrspace(1) %out
|
|
; CHECK-NEXT: ret void
|
|
; CHECK-NEXT: }
|
|
bb:
|
|
%res = call <8 x float> @llvm.amdgcn.wmma.f32.16x16x32.f16.v8f32.v16f16(i1 0, <16 x half> %A, i1 0, <16 x half> %B, i16 0, <8 x float> %C, i1 false, i1 true)
|
|
store <8 x float> %res, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_ps void @test_wmma_f16_16x16x32_f16(<16 x half> %A, <16 x half> %B, <8 x half> %C, ptr addrspace(1) %out) {
|
|
; CHECK-LABEL: define amdgpu_ps void @test_wmma_f16_16x16x32_f16(<16 x half> %A, <16 x half> %B, <8 x half> %C, ptr addrspace(1) %out) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: %res = 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-NEXT: store <8 x half> %res, ptr addrspace(1) %out
|
|
; CHECK-NEXT: ret void
|
|
; CHECK-NEXT: }
|
|
bb:
|
|
%res = call <8 x half> @llvm.amdgcn.wmma.f16.16x16x32.f16.v8f16.v16f16(i1 0, <16 x half> %A, i1 0, <16 x half> %B, i16 0, <8 x half> %C, i1 false, i1 true)
|
|
store <8 x half> %res, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_ps void @test_wmma_bf16f32_16x16x32_bf16(<16 x bfloat> %A, <16 x bfloat> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-LABEL: define amdgpu_ps void @test_wmma_bf16f32_16x16x32_bf16(<16 x bfloat> %A, <16 x bfloat> %B, <8 x float> %C, ptr addrspace(1) %out) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: %res = 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-NEXT: store <8 x bfloat> %res, ptr addrspace(1) %out
|
|
; CHECK-NEXT: ret void
|
|
; CHECK-NEXT: }
|
|
bb:
|
|
%res = call <8 x bfloat> @llvm.amdgcn.wmma.bf16f32.16x16x32.bf16.v8bf16.v16bf16(i1 0, <16 x bfloat> %A, i1 0, <16 x bfloat> %B, i16 0, <8 x float> %C, i1 false, i1 true)
|
|
store <8 x bfloat> %res, ptr addrspace(1) %out
|
|
ret void
|
|
}
|