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
85 lines
4.2 KiB
LLVM
85 lines
4.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt < %s -passes=amdgpu-lower-intrinsics -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 -codegen-opt-level=0 | FileCheck --check-prefixes=GFX11,GFX11-NOOPT %s
|
|
; RUN: opt < %s -passes=amdgpu-lower-intrinsics -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 -codegen-opt-level=1 -mattr=+wavefrontsize32 | FileCheck --check-prefixes=GFX11,OPT-WAVE32,GFX11-OPT-WAVE32 %s
|
|
; RUN: opt < %s -passes=amdgpu-lower-intrinsics -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 -codegen-opt-level=1 -mattr=+wavefrontsize64 | FileCheck --check-prefixes=GFX11,OPT-WAVE64,GFX11-OPT-WAVE64 %s
|
|
; RUN: opt < %s -passes=amdgpu-lower-intrinsics -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1200 -codegen-opt-level=0 | FileCheck --check-prefixes=GFX12,GFX12-NOOPT %s
|
|
; RUN: opt < %s -passes=amdgpu-lower-intrinsics -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1200 -codegen-opt-level=1 -mattr=+wavefrontsize32 | FileCheck --check-prefixes=GFX12,OPT-WAVE32,GFX12-OPT-WAVE32 %s
|
|
; RUN: opt < %s -passes=amdgpu-lower-intrinsics -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1200 -codegen-opt-level=1 -mattr=+wavefrontsize64 | FileCheck --check-prefixes=GFX12,OPT-WAVE64,GFX12-OPT-WAVE64 %s
|
|
|
|
define amdgpu_kernel void @barrier() {
|
|
; GFX11-LABEL: define amdgpu_kernel void @barrier(
|
|
; GFX11-SAME: ) #[[ATTR0:[0-9]+]] {
|
|
; GFX11-NEXT: call void @llvm.amdgcn.s.barrier()
|
|
; GFX11-NEXT: ret void
|
|
;
|
|
; GFX12-LABEL: define amdgpu_kernel void @barrier(
|
|
; GFX12-SAME: ) #[[ATTR0:[0-9]+]] {
|
|
; GFX12-NEXT: call void @llvm.amdgcn.s.barrier.signal(i32 -1)
|
|
; GFX12-NEXT: call void @llvm.amdgcn.s.barrier.wait(i16 -1)
|
|
; GFX12-NEXT: ret void
|
|
;
|
|
call void @llvm.amdgcn.s.barrier()
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @barrier_32threads() "amdgpu-flat-work-group-size"="32,32" {
|
|
; GFX11-NOOPT-LABEL: define amdgpu_kernel void @barrier_32threads(
|
|
; GFX11-NOOPT-SAME: ) #[[ATTR1:[0-9]+]] {
|
|
; GFX11-NOOPT-NEXT: call void @llvm.amdgcn.s.barrier()
|
|
; GFX11-NOOPT-NEXT: ret void
|
|
;
|
|
; OPT-WAVE32-LABEL: define amdgpu_kernel void @barrier_32threads(
|
|
; OPT-WAVE32-SAME: ) #[[ATTR1:[0-9]+]] {
|
|
; OPT-WAVE32-NEXT: call void @llvm.amdgcn.wave.barrier()
|
|
; OPT-WAVE32-NEXT: ret void
|
|
;
|
|
; OPT-WAVE64-LABEL: define amdgpu_kernel void @barrier_32threads(
|
|
; OPT-WAVE64-SAME: ) #[[ATTR1:[0-9]+]] {
|
|
; OPT-WAVE64-NEXT: call void @llvm.amdgcn.wave.barrier()
|
|
; OPT-WAVE64-NEXT: ret void
|
|
;
|
|
; GFX12-NOOPT-LABEL: define amdgpu_kernel void @barrier_32threads(
|
|
; GFX12-NOOPT-SAME: ) #[[ATTR1:[0-9]+]] {
|
|
; GFX12-NOOPT-NEXT: call void @llvm.amdgcn.s.barrier.signal(i32 -1)
|
|
; GFX12-NOOPT-NEXT: call void @llvm.amdgcn.s.barrier.wait(i16 -1)
|
|
; GFX12-NOOPT-NEXT: ret void
|
|
;
|
|
call void @llvm.amdgcn.s.barrier()
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @barrier_64threads() "amdgpu-flat-work-group-size"="64,64" {
|
|
; GFX11-NOOPT-LABEL: define amdgpu_kernel void @barrier_64threads(
|
|
; GFX11-NOOPT-SAME: ) #[[ATTR2:[0-9]+]] {
|
|
; GFX11-NOOPT-NEXT: call void @llvm.amdgcn.s.barrier()
|
|
; GFX11-NOOPT-NEXT: ret void
|
|
;
|
|
; GFX11-OPT-WAVE32-LABEL: define amdgpu_kernel void @barrier_64threads(
|
|
; GFX11-OPT-WAVE32-SAME: ) #[[ATTR2:[0-9]+]] {
|
|
; GFX11-OPT-WAVE32-NEXT: call void @llvm.amdgcn.s.barrier()
|
|
; GFX11-OPT-WAVE32-NEXT: ret void
|
|
;
|
|
; OPT-WAVE64-LABEL: define amdgpu_kernel void @barrier_64threads(
|
|
; OPT-WAVE64-SAME: ) #[[ATTR2:[0-9]+]] {
|
|
; OPT-WAVE64-NEXT: call void @llvm.amdgcn.wave.barrier()
|
|
; OPT-WAVE64-NEXT: ret void
|
|
;
|
|
; GFX12-NOOPT-LABEL: define amdgpu_kernel void @barrier_64threads(
|
|
; GFX12-NOOPT-SAME: ) #[[ATTR2:[0-9]+]] {
|
|
; GFX12-NOOPT-NEXT: call void @llvm.amdgcn.s.barrier.signal(i32 -1)
|
|
; GFX12-NOOPT-NEXT: call void @llvm.amdgcn.s.barrier.wait(i16 -1)
|
|
; GFX12-NOOPT-NEXT: ret void
|
|
;
|
|
; GFX12-OPT-WAVE32-LABEL: define amdgpu_kernel void @barrier_64threads(
|
|
; GFX12-OPT-WAVE32-SAME: ) #[[ATTR2:[0-9]+]] {
|
|
; GFX12-OPT-WAVE32-NEXT: call void @llvm.amdgcn.s.barrier.signal(i32 -1)
|
|
; GFX12-OPT-WAVE32-NEXT: call void @llvm.amdgcn.s.barrier.wait(i16 -1)
|
|
; GFX12-OPT-WAVE32-NEXT: ret void
|
|
;
|
|
call void @llvm.amdgcn.s.barrier()
|
|
ret void
|
|
}
|
|
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
|
|
; GFX11-OPT-WAVE64: {{.*}}
|
|
; GFX12-OPT-WAVE64: {{.*}}
|