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
215 lines
4.4 KiB
LLVM
215 lines
4.4 KiB
LLVM
; RUN: llc -mtriple=amdgcn-- -mcpu=verde < %s | FileCheck -check-prefixes=GCN,SICIVI,FUNC %s
|
|
; RUN: llc -mtriple=amdgcn-- -mcpu=tonga < %s | FileCheck -check-prefixes=GCN,SICIVI,VI,FUNC %s
|
|
; RUN: llc -mtriple=amdgcn-- -mcpu=gfx900 < %s | FileCheck -check-prefixes=GCN,GFX9,FUNC %s
|
|
; RUN: llc -mtriple=r600-- -mcpu=redwood < %s | FileCheck -check-prefixes=EG,FUNC %s
|
|
; RUN: llc -mtriple=r600-- -mcpu=cayman < %s | FileCheck -check-prefixes=CM,FUNC %s
|
|
|
|
; FUNC-LABEL: {{^}}store_local_i1:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_BYTE_WRITE
|
|
|
|
; CM: LDS_BYTE_WRITE
|
|
|
|
; GCN: ds_write_b8
|
|
define amdgpu_kernel void @store_local_i1(ptr addrspace(3) %out) {
|
|
entry:
|
|
store i1 true, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_i8:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_BYTE_WRITE
|
|
|
|
; CM: LDS_BYTE_WRITE
|
|
|
|
; GCN: ds_write_b8
|
|
define amdgpu_kernel void @store_local_i8(ptr addrspace(3) %out, i8 %in) {
|
|
store i8 %in, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_i16:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_SHORT_WRITE
|
|
|
|
; CM: LDS_SHORT_WRITE
|
|
|
|
; GCN: ds_write_b16
|
|
define amdgpu_kernel void @store_local_i16(ptr addrspace(3) %out, i16 %in) {
|
|
store i16 %in, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_v2i16:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_WRITE
|
|
|
|
; CM: LDS_WRITE
|
|
|
|
; GCN: ds_write_b32
|
|
define amdgpu_kernel void @store_local_v2i16(ptr addrspace(3) %out, <2 x i16> %in) {
|
|
entry:
|
|
store <2 x i16> %in, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_v4i8:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_WRITE
|
|
|
|
; CM: LDS_WRITE
|
|
|
|
; GCN: ds_write_b32
|
|
define amdgpu_kernel void @store_local_v4i8(ptr addrspace(3) %out, <4 x i8> %in) {
|
|
entry:
|
|
store <4 x i8> %in, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_v4i8_unaligned:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_BYTE_WRITE
|
|
; EG: LDS_BYTE_WRITE
|
|
; EG: LDS_BYTE_WRITE
|
|
; EG: LDS_BYTE_WRITE
|
|
; EG-NOT: LDS_WRITE
|
|
|
|
; CM: LDS_BYTE_WRITE
|
|
; CM: LDS_BYTE_WRITE
|
|
; CM: LDS_BYTE_WRITE
|
|
; CM: LDS_BYTE_WRITE
|
|
; CM-NOT: LDS_WRITE
|
|
|
|
; GCN: ds_write_b8
|
|
; GCN: ds_write_b8
|
|
; GCN: ds_write_b8
|
|
; GCN: ds_write_b8
|
|
define amdgpu_kernel void @store_local_v4i8_unaligned(ptr addrspace(3) %out, <4 x i8> %in) {
|
|
entry:
|
|
store <4 x i8> %in, ptr addrspace(3) %out, align 1
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_v4i8_halfaligned:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_SHORT_WRITE
|
|
; EG: LDS_SHORT_WRITE
|
|
; EG-NOT: LDS_WRITE
|
|
|
|
; CM: LDS_SHORT_WRITE
|
|
; CM: LDS_SHORT_WRITE
|
|
; CM-NOT: LDS_WRITE
|
|
|
|
; GCN: ds_write_b16
|
|
; GCN: ds_write_b16
|
|
define amdgpu_kernel void @store_local_v4i8_halfaligned(ptr addrspace(3) %out, <4 x i8> %in) {
|
|
entry:
|
|
store <4 x i8> %in, ptr addrspace(3) %out, align 2
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_v2i32:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_WRITE
|
|
; EG: LDS_WRITE
|
|
; EG-NOT: LDS_WRITE
|
|
|
|
; CM: LDS_WRITE
|
|
; CM: LDS_WRITE
|
|
; CM-NOT: LDS_WRITE
|
|
|
|
; GCN: ds_write_b64
|
|
define amdgpu_kernel void @store_local_v2i32(ptr addrspace(3) %out, <2 x i32> %in) {
|
|
entry:
|
|
store <2 x i32> %in, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_v4i32:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_WRITE
|
|
; EG: LDS_WRITE
|
|
; EG: LDS_WRITE
|
|
; EG: LDS_WRITE
|
|
|
|
; CM: LDS_WRITE
|
|
; CM: LDS_WRITE
|
|
; CM: LDS_WRITE
|
|
; CM: LDS_WRITE
|
|
|
|
; SI: ds_write2_b32
|
|
; VI: ds_write_b128
|
|
; GFX9: ds_write_b128
|
|
define amdgpu_kernel void @store_local_v4i32(ptr addrspace(3) %out, <4 x i32> %in) {
|
|
entry:
|
|
store <4 x i32> %in, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_v4i32_align4:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_WRITE
|
|
; EG: LDS_WRITE
|
|
; EG: LDS_WRITE
|
|
; EG: LDS_WRITE
|
|
|
|
; CM: LDS_WRITE
|
|
; CM: LDS_WRITE
|
|
; CM: LDS_WRITE
|
|
; CM: LDS_WRITE
|
|
|
|
; GCN: ds_write2_b32
|
|
; GCN: ds_write2_b32
|
|
define amdgpu_kernel void @store_local_v4i32_align4(ptr addrspace(3) %out, <4 x i32> %in) {
|
|
entry:
|
|
store <4 x i32> %in, ptr addrspace(3) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_i64_i8:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_BYTE_WRITE
|
|
; GCN: ds_write_b8
|
|
define amdgpu_kernel void @store_local_i64_i8(ptr addrspace(3) %out, i64 %in) {
|
|
entry:
|
|
%0 = trunc i64 %in to i8
|
|
store i8 %0, ptr addrspace(3) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}store_local_i64_i16:
|
|
; SICIVI: s_mov_b32 m0
|
|
; GFX9-NOT: m0
|
|
|
|
; EG: LDS_SHORT_WRITE
|
|
; GCN: ds_write_b16
|
|
define amdgpu_kernel void @store_local_i64_i16(ptr addrspace(3) %out, i64 %in) {
|
|
entry:
|
|
%0 = trunc i64 %in to i16
|
|
store i16 %0, ptr addrspace(3) %out
|
|
ret void
|
|
}
|