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
53 lines
2.5 KiB
LLVM
53 lines
2.5 KiB
LLVM
; RUN: llc -mtriple=amdgcn--amdpal -mcpu=tonga < %s | FileCheck -check-prefixes=GCN %s
|
|
|
|
; The first image store and the second image load use the same descriptor and
|
|
; the same coordinate. Check that they do not get swapped by the machine
|
|
; instruction scheduler.
|
|
|
|
; GCN-LABEL: {{^}}_amdgpu_cs_main:
|
|
; GCN: image_load
|
|
; GCN: image_store
|
|
; GCN: image_load
|
|
; GCN: image_store
|
|
|
|
define dllexport amdgpu_cs void @_amdgpu_cs_main(i32 inreg %arg, i32 inreg %arg1, i32 inreg %arg2, <3 x i32> inreg %arg3, i32 inreg %arg4, <3 x i32> %arg5) local_unnamed_addr #0 {
|
|
.entry:
|
|
%tmp = call i64 @llvm.amdgcn.s.getpc() #1
|
|
%tmp6 = bitcast i64 %tmp to <2 x i32>
|
|
%.0.vec.insert = insertelement <2 x i32> poison, i32 %arg2, i32 0
|
|
%.4.vec.insert = shufflevector <2 x i32> %.0.vec.insert, <2 x i32> %tmp6, <2 x i32> <i32 0, i32 3>
|
|
%tmp7 = bitcast <2 x i32> %.4.vec.insert to i64
|
|
%tmp8 = inttoptr i64 %tmp7 to ptr addrspace(4)
|
|
%tmp9 = add <3 x i32> %arg3, %arg5
|
|
%tmp10 = getelementptr [4294967295 x i8], ptr addrspace(4) %tmp8, i64 0, i64 32
|
|
%tmp12 = load <8 x i32>, ptr addrspace(4) %tmp10, align 16
|
|
%tmp13.0 = extractelement <3 x i32> %tmp9, i32 0
|
|
%tmp13.1 = extractelement <3 x i32> %tmp9, i32 1
|
|
%tmp14 = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 15, i32 %tmp13.0, i32 %tmp13.1, <8 x i32> %tmp12, i32 0, i32 0) #0
|
|
%tmp15 = inttoptr i64 %tmp7 to ptr addrspace(4)
|
|
%tmp16 = load <8 x i32>, ptr addrspace(4) %tmp15, align 16
|
|
call void @llvm.amdgcn.image.store.2d.v4f32.i32(<4 x float> %tmp14, i32 15, i32 %tmp13.0, i32 %tmp13.1, <8 x i32> %tmp16, i32 0, i32 0) #0
|
|
%tmp17 = load <8 x i32>, ptr addrspace(4) %tmp15, align 16
|
|
%tmp18 = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 165, i32 %tmp13.0, i32 %tmp13.1, <8 x i32> %tmp17, i32 0, i32 0) #0
|
|
%tmp19 = getelementptr [4294967295 x i8], ptr addrspace(4) %tmp8, i64 0, i64 64
|
|
%tmp21 = load <8 x i32>, ptr addrspace(4) %tmp19, align 16
|
|
call void @llvm.amdgcn.image.store.2d.v4f32.i32(<4 x float> %tmp18, i32 15, i32 %tmp13.0, i32 %tmp13.1, <8 x i32> %tmp21, i32 0, i32 0) #0
|
|
ret void
|
|
}
|
|
|
|
; Function Attrs: nounwind readnone speculatable
|
|
declare i64 @llvm.amdgcn.s.getpc() #1
|
|
|
|
; Function Attrs: nounwind readonly
|
|
declare <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32, i32, i32, <8 x i32>, i32, i32) #2
|
|
|
|
; Function Attrs: nounwind writeonly
|
|
declare void @llvm.amdgcn.image.store.2d.v4f32.i32(<4 x float>, i32, i32, i32, <8 x i32>, i32, i32) #3
|
|
|
|
attributes #0 = { nounwind }
|
|
attributes #1 = { nounwind readnone speculatable }
|
|
attributes #2 = { nounwind readonly }
|
|
attributes #3 = { nounwind writeonly }
|
|
|
|
!0 = !{}
|