Files
TheRock/compiler/amd-llvm/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ds.gws.barrier-bundle.ll
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
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

173 lines
10 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 6
; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -stop-after=postrapseudos < %s | FileCheck -check-prefix=GFX6-SDAG %s
; RUN: llc -global-isel=1 -new-reg-bank-select -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -stop-after=postrapseudos < %s | FileCheck -check-prefix=GFX6-GISEL %s
; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -stop-after=postrapseudos < %s | FileCheck -check-prefix=GFX9-SDAG %s
; RUN: llc -global-isel=1 -new-reg-bank-select -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -stop-after=postrapseudos < %s | FileCheck -check-prefix=GFX9-GISEL %s
; Make sure the op is emitted bundled with a waitcnt with and without the retry loop, and the bundle is not removed by ExpandPostRAPseudos.
define amdgpu_kernel void @gws_barrier_offset0(i32 %val) #0 {
; GFX6-SDAG-LABEL: name: gws_barrier_offset0
; GFX6-SDAG: bb.0 (%ir-block.0):
; GFX6-SDAG-NEXT: successors: %bb.1(0x80000000)
; GFX6-SDAG-NEXT: liveins: $sgpr8_sgpr9
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX6-SDAG-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX6-SDAG-NEXT: $m0 = S_MOV_B32 0
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: bb.1:
; GFX6-SDAG-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
; GFX6-SDAG-NEXT: liveins: $vgpr0
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: S_SETREG_IMM32_B32 0, 515, implicit-def $mode, implicit $mode
; GFX6-SDAG-NEXT: BUNDLE implicit renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX6-SDAG-NEXT: DS_GWS_BARRIER renamable $vgpr0, 0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX6-SDAG-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX6-SDAG-NEXT: }
; GFX6-SDAG-NEXT: renamable $sgpr0 = S_GETREG_B32 515, implicit $mode
; GFX6-SDAG-NEXT: S_CMP_LG_U32 killed renamable $sgpr0, 0, implicit-def $scc
; GFX6-SDAG-NEXT: S_CBRANCH_SCC1 %bb.1, implicit killed $scc
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: bb.2:
; GFX6-SDAG-NEXT: S_ENDPGM 0
;
; GFX6-GISEL-LABEL: name: gws_barrier_offset0
; GFX6-GISEL: bb.0 (%ir-block.0):
; GFX6-GISEL-NEXT: successors: %bb.1(0x80000000)
; GFX6-GISEL-NEXT: liveins: $sgpr8_sgpr9
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX6-GISEL-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX6-GISEL-NEXT: $m0 = S_MOV_B32 0
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: bb.1:
; GFX6-GISEL-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
; GFX6-GISEL-NEXT: liveins: $vgpr0
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: S_SETREG_IMM32_B32 0, 515, implicit-def $mode, implicit $mode
; GFX6-GISEL-NEXT: BUNDLE implicit renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX6-GISEL-NEXT: DS_GWS_BARRIER renamable $vgpr0, 0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX6-GISEL-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX6-GISEL-NEXT: }
; GFX6-GISEL-NEXT: renamable $sgpr0 = S_GETREG_B32 515, implicit $mode
; GFX6-GISEL-NEXT: S_CMP_LG_U32 killed renamable $sgpr0, 0, implicit-def $scc
; GFX6-GISEL-NEXT: S_CBRANCH_SCC1 %bb.1, implicit killed $scc
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: bb.2:
; GFX6-GISEL-NEXT: S_ENDPGM 0
;
; GFX9-SDAG-LABEL: name: gws_barrier_offset0
; GFX9-SDAG: bb.0 (%ir-block.0):
; GFX9-SDAG-NEXT: liveins: $sgpr8_sgpr9
; GFX9-SDAG-NEXT: {{ $}}
; GFX9-SDAG-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX9-SDAG-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX9-SDAG-NEXT: $m0 = S_MOV_B32 0
; GFX9-SDAG-NEXT: BUNDLE implicit killed renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX9-SDAG-NEXT: DS_GWS_BARRIER renamable $vgpr0, 0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX9-SDAG-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX9-SDAG-NEXT: }
; GFX9-SDAG-NEXT: S_ENDPGM 0
;
; GFX9-GISEL-LABEL: name: gws_barrier_offset0
; GFX9-GISEL: bb.0 (%ir-block.0):
; GFX9-GISEL-NEXT: liveins: $sgpr8_sgpr9
; GFX9-GISEL-NEXT: {{ $}}
; GFX9-GISEL-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX9-GISEL-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX9-GISEL-NEXT: $m0 = S_MOV_B32 0
; GFX9-GISEL-NEXT: BUNDLE implicit killed renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX9-GISEL-NEXT: DS_GWS_BARRIER renamable $vgpr0, 0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX9-GISEL-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX9-GISEL-NEXT: }
; GFX9-GISEL-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.ds.gws.barrier(i32 %val, i32 0)
ret void
}
define amdgpu_kernel void @gws_barrier_offset63(i32 %val) #0 {
; GFX6-SDAG-LABEL: name: gws_barrier_offset63
; GFX6-SDAG: bb.0 (%ir-block.0):
; GFX6-SDAG-NEXT: successors: %bb.1(0x80000000)
; GFX6-SDAG-NEXT: liveins: $sgpr8_sgpr9
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX6-SDAG-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX6-SDAG-NEXT: $m0 = S_MOV_B32 0
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: bb.1:
; GFX6-SDAG-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
; GFX6-SDAG-NEXT: liveins: $vgpr0
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: S_SETREG_IMM32_B32 0, 515, implicit-def $mode, implicit $mode
; GFX6-SDAG-NEXT: BUNDLE implicit renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX6-SDAG-NEXT: DS_GWS_BARRIER renamable $vgpr0, 63, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX6-SDAG-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX6-SDAG-NEXT: }
; GFX6-SDAG-NEXT: renamable $sgpr0 = S_GETREG_B32 515, implicit $mode
; GFX6-SDAG-NEXT: S_CMP_LG_U32 killed renamable $sgpr0, 0, implicit-def $scc
; GFX6-SDAG-NEXT: S_CBRANCH_SCC1 %bb.1, implicit killed $scc
; GFX6-SDAG-NEXT: {{ $}}
; GFX6-SDAG-NEXT: bb.2:
; GFX6-SDAG-NEXT: S_ENDPGM 0
;
; GFX6-GISEL-LABEL: name: gws_barrier_offset63
; GFX6-GISEL: bb.0 (%ir-block.0):
; GFX6-GISEL-NEXT: successors: %bb.1(0x80000000)
; GFX6-GISEL-NEXT: liveins: $sgpr8_sgpr9
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX6-GISEL-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX6-GISEL-NEXT: $m0 = S_MOV_B32 0
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: bb.1:
; GFX6-GISEL-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
; GFX6-GISEL-NEXT: liveins: $vgpr0
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: S_SETREG_IMM32_B32 0, 515, implicit-def $mode, implicit $mode
; GFX6-GISEL-NEXT: BUNDLE implicit renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX6-GISEL-NEXT: DS_GWS_BARRIER renamable $vgpr0, 63, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX6-GISEL-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX6-GISEL-NEXT: }
; GFX6-GISEL-NEXT: renamable $sgpr0 = S_GETREG_B32 515, implicit $mode
; GFX6-GISEL-NEXT: S_CMP_LG_U32 killed renamable $sgpr0, 0, implicit-def $scc
; GFX6-GISEL-NEXT: S_CBRANCH_SCC1 %bb.1, implicit killed $scc
; GFX6-GISEL-NEXT: {{ $}}
; GFX6-GISEL-NEXT: bb.2:
; GFX6-GISEL-NEXT: S_ENDPGM 0
;
; GFX9-SDAG-LABEL: name: gws_barrier_offset63
; GFX9-SDAG: bb.0 (%ir-block.0):
; GFX9-SDAG-NEXT: liveins: $sgpr8_sgpr9
; GFX9-SDAG-NEXT: {{ $}}
; GFX9-SDAG-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX9-SDAG-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX9-SDAG-NEXT: $m0 = S_MOV_B32 0
; GFX9-SDAG-NEXT: BUNDLE implicit killed renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX9-SDAG-NEXT: DS_GWS_BARRIER renamable $vgpr0, 63, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX9-SDAG-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX9-SDAG-NEXT: }
; GFX9-SDAG-NEXT: S_ENDPGM 0
;
; GFX9-GISEL-LABEL: name: gws_barrier_offset63
; GFX9-GISEL: bb.0 (%ir-block.0):
; GFX9-GISEL-NEXT: liveins: $sgpr8_sgpr9
; GFX9-GISEL-NEXT: {{ $}}
; GFX9-GISEL-NEXT: renamable $sgpr0 = S_LOAD_DWORD_IMM killed renamable $sgpr8_sgpr9, 0, 0 :: (dereferenceable invariant load (s32) from %ir.val.kernarg.offset1, align 16, addrspace 4)
; GFX9-GISEL-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
; GFX9-GISEL-NEXT: $m0 = S_MOV_B32 0
; GFX9-GISEL-NEXT: BUNDLE implicit killed renamable $vgpr0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") {
; GFX9-GISEL-NEXT: DS_GWS_BARRIER renamable $vgpr0, 63, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
; GFX9-GISEL-NEXT: S_WAITCNT .Vmcnt_0_Expcnt_0_Lgkmcnt_0
; GFX9-GISEL-NEXT: }
; GFX9-GISEL-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.ds.gws.barrier(i32 %val, i32 63)
ret void
}
declare void @llvm.amdgcn.ds.gws.barrier(i32, i32) #1
attributes #0 = { nounwind }
attributes #1 = { convergent inaccessiblememonly nounwind }