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
221 lines
8.0 KiB
LLVM
221 lines
8.0 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx700 < %s | FileCheck -check-prefix=GFX7 %s
|
|
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx803 < %s | FileCheck -check-prefix=GFX8 %s
|
|
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 < %s | FileCheck -check-prefix=GFX900 %s
|
|
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1010 < %s | FileCheck -check-prefix=GFX10 %s
|
|
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx942 < %s | FileCheck -check-prefix=GFX940 %s
|
|
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 < %s | FileCheck -check-prefix=GFX11 %s
|
|
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1200 < %s | FileCheck -check-prefix=GFX12 %s
|
|
|
|
|
|
define void @s_add_use_fi_sgpr_offset(i32 inreg %soffset) #0 {
|
|
; GFX7-LABEL: s_add_use_fi_sgpr_offset:
|
|
; GFX7: ; %bb.0:
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b32 s7, s33
|
|
; GFX7-NEXT: s_add_i32 s33, s32, 0x7c0
|
|
; GFX7-NEXT: s_and_b32 s33, s33, 0xfffff800
|
|
; GFX7-NEXT: s_mov_b32 s8, s34
|
|
; GFX7-NEXT: s_mov_b32 s34, s32
|
|
; GFX7-NEXT: s_add_i32 s32, s32, 0x45000
|
|
; GFX7-NEXT: s_and_b32 s4, s16, 31
|
|
; GFX7-NEXT: s_lshr_b32 s5, s33, 6
|
|
; GFX7-NEXT: s_addk_i32 s5, 0x120
|
|
; GFX7-NEXT: s_lshr_b32 s6, s33, 6
|
|
; GFX7-NEXT: s_add_i32 s6, s6, 16
|
|
; GFX7-NEXT: ;;#ASMSTART
|
|
; GFX7-NEXT: ; use s5
|
|
; GFX7-NEXT: ;;#ASMEND
|
|
; GFX7-NEXT: s_lshr_b32 s5, s33, 6
|
|
; GFX7-NEXT: s_addk_i32 s5, 0x120
|
|
; GFX7-NEXT: s_or_b32 s4, s4, s5
|
|
; GFX7-NEXT: ;;#ASMSTART
|
|
; GFX7-NEXT: ; use s4
|
|
; GFX7-NEXT: ;;#ASMEND
|
|
; GFX7-NEXT: s_mov_b32 s32, s34
|
|
; GFX7-NEXT: s_mov_b32 s34, s8
|
|
; GFX7-NEXT: s_mov_b32 s33, s7
|
|
; GFX7-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX8-LABEL: s_add_use_fi_sgpr_offset:
|
|
; GFX8: ; %bb.0:
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX8-NEXT: s_mov_b32 s7, s33
|
|
; GFX8-NEXT: s_add_i32 s33, s32, 0x7c0
|
|
; GFX8-NEXT: s_and_b32 s33, s33, 0xfffff800
|
|
; GFX8-NEXT: s_mov_b32 s8, s34
|
|
; GFX8-NEXT: s_mov_b32 s34, s32
|
|
; GFX8-NEXT: s_add_i32 s32, s32, 0x45000
|
|
; GFX8-NEXT: s_and_b32 s4, s16, 31
|
|
; GFX8-NEXT: s_lshr_b32 s5, s33, 6
|
|
; GFX8-NEXT: s_addk_i32 s5, 0x120
|
|
; GFX8-NEXT: s_lshr_b32 s6, s33, 6
|
|
; GFX8-NEXT: s_add_i32 s6, s6, 16
|
|
; GFX8-NEXT: ;;#ASMSTART
|
|
; GFX8-NEXT: ; use s5
|
|
; GFX8-NEXT: ;;#ASMEND
|
|
; GFX8-NEXT: s_lshr_b32 s5, s33, 6
|
|
; GFX8-NEXT: s_addk_i32 s5, 0x120
|
|
; GFX8-NEXT: s_or_b32 s4, s4, s5
|
|
; GFX8-NEXT: ;;#ASMSTART
|
|
; GFX8-NEXT: ; use s4
|
|
; GFX8-NEXT: ;;#ASMEND
|
|
; GFX8-NEXT: s_mov_b32 s32, s34
|
|
; GFX8-NEXT: s_mov_b32 s34, s8
|
|
; GFX8-NEXT: s_mov_b32 s33, s7
|
|
; GFX8-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX900-LABEL: s_add_use_fi_sgpr_offset:
|
|
; GFX900: ; %bb.0:
|
|
; GFX900-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX900-NEXT: s_mov_b32 s7, s33
|
|
; GFX900-NEXT: s_add_i32 s33, s32, 0x7c0
|
|
; GFX900-NEXT: s_and_b32 s33, s33, 0xfffff800
|
|
; GFX900-NEXT: s_mov_b32 s8, s34
|
|
; GFX900-NEXT: s_mov_b32 s34, s32
|
|
; GFX900-NEXT: s_add_i32 s32, s32, 0x45000
|
|
; GFX900-NEXT: s_and_b32 s4, s16, 31
|
|
; GFX900-NEXT: s_lshr_b32 s5, s33, 6
|
|
; GFX900-NEXT: s_addk_i32 s5, 0x120
|
|
; GFX900-NEXT: s_lshr_b32 s6, s33, 6
|
|
; GFX900-NEXT: s_add_i32 s6, s6, 16
|
|
; GFX900-NEXT: ;;#ASMSTART
|
|
; GFX900-NEXT: ; use s5
|
|
; GFX900-NEXT: ;;#ASMEND
|
|
; GFX900-NEXT: s_lshr_b32 s5, s33, 6
|
|
; GFX900-NEXT: s_addk_i32 s5, 0x120
|
|
; GFX900-NEXT: s_or_b32 s4, s4, s5
|
|
; GFX900-NEXT: ;;#ASMSTART
|
|
; GFX900-NEXT: ; use s4
|
|
; GFX900-NEXT: ;;#ASMEND
|
|
; GFX900-NEXT: s_mov_b32 s32, s34
|
|
; GFX900-NEXT: s_mov_b32 s34, s8
|
|
; GFX900-NEXT: s_mov_b32 s33, s7
|
|
; GFX900-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX10-LABEL: s_add_use_fi_sgpr_offset:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: s_mov_b32 s7, s33
|
|
; GFX10-NEXT: s_add_i32 s33, s32, 0x3e0
|
|
; GFX10-NEXT: s_mov_b32 s8, s34
|
|
; GFX10-NEXT: s_and_b32 s33, s33, 0xfffffc00
|
|
; GFX10-NEXT: s_mov_b32 s34, s32
|
|
; GFX10-NEXT: s_add_i32 s32, s32, 0x22800
|
|
; GFX10-NEXT: s_and_b32 s4, s16, 31
|
|
; GFX10-NEXT: s_lshr_b32 s5, s33, 5
|
|
; GFX10-NEXT: s_mov_b32 s32, s34
|
|
; GFX10-NEXT: s_addk_i32 s5, 0x120
|
|
; GFX10-NEXT: s_lshr_b32 s6, s33, 5
|
|
; GFX10-NEXT: s_mov_b32 s34, s8
|
|
; GFX10-NEXT: s_add_i32 s6, s6, 16
|
|
; GFX10-NEXT: s_lshr_b32 s9, s33, 5
|
|
; GFX10-NEXT: ;;#ASMSTART
|
|
; GFX10-NEXT: ; use s5
|
|
; GFX10-NEXT: ;;#ASMEND
|
|
; GFX10-NEXT: s_addk_i32 s9, 0x120
|
|
; GFX10-NEXT: s_or_b32 s4, s4, s9
|
|
; GFX10-NEXT: ;;#ASMSTART
|
|
; GFX10-NEXT: ; use s4
|
|
; GFX10-NEXT: ;;#ASMEND
|
|
; GFX10-NEXT: s_mov_b32 s33, s7
|
|
; GFX10-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX940-LABEL: s_add_use_fi_sgpr_offset:
|
|
; GFX940: ; %bb.0:
|
|
; GFX940-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX940-NEXT: s_mov_b32 s3, s33
|
|
; GFX940-NEXT: s_add_i32 s33, s32, 31
|
|
; GFX940-NEXT: s_andn2_b32 s33, s33, 31
|
|
; GFX940-NEXT: s_mov_b32 s4, s34
|
|
; GFX940-NEXT: s_mov_b32 s34, s32
|
|
; GFX940-NEXT: s_addk_i32 s32, 0x1140
|
|
; GFX940-NEXT: s_and_b32 s0, s0, 31
|
|
; GFX940-NEXT: s_add_i32 s2, s33, 0x120
|
|
; GFX940-NEXT: s_mov_b32 s1, s2
|
|
; GFX940-NEXT: s_add_i32 s5, s33, 16
|
|
; GFX940-NEXT: s_mov_b32 s2, s5
|
|
; GFX940-NEXT: ;;#ASMSTART
|
|
; GFX940-NEXT: ; use s1
|
|
; GFX940-NEXT: ;;#ASMEND
|
|
; GFX940-NEXT: s_add_i32 s1, s33, 0x120
|
|
; GFX940-NEXT: s_or_b32 s0, s0, s1
|
|
; GFX940-NEXT: ;;#ASMSTART
|
|
; GFX940-NEXT: ; use s0
|
|
; GFX940-NEXT: ;;#ASMEND
|
|
; GFX940-NEXT: s_mov_b32 s32, s34
|
|
; GFX940-NEXT: s_mov_b32 s34, s4
|
|
; GFX940-NEXT: s_mov_b32 s33, s3
|
|
; GFX940-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX11-LABEL: s_add_use_fi_sgpr_offset:
|
|
; GFX11: ; %bb.0:
|
|
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX11-NEXT: s_mov_b32 s3, s33
|
|
; GFX11-NEXT: s_add_i32 s33, s32, 31
|
|
; GFX11-NEXT: s_mov_b32 s4, s34
|
|
; GFX11-NEXT: s_and_not1_b32 s33, s33, 31
|
|
; GFX11-NEXT: s_mov_b32 s34, s32
|
|
; GFX11-NEXT: s_addk_i32 s32, 0x1140
|
|
; GFX11-NEXT: s_and_b32 s0, s0, 31
|
|
; GFX11-NEXT: s_add_i32 s2, s33, 0x120
|
|
; GFX11-NEXT: s_add_i32 s5, s33, 16
|
|
; GFX11-NEXT: s_mov_b32 s1, s2
|
|
; GFX11-NEXT: s_mov_b32 s2, s5
|
|
; GFX11-NEXT: s_add_i32 s5, s33, 0x120
|
|
; GFX11-NEXT: ;;#ASMSTART
|
|
; GFX11-NEXT: ; use s1
|
|
; GFX11-NEXT: ;;#ASMEND
|
|
; GFX11-NEXT: s_or_b32 s0, s0, s5
|
|
; GFX11-NEXT: ;;#ASMSTART
|
|
; GFX11-NEXT: ; use s0
|
|
; GFX11-NEXT: ;;#ASMEND
|
|
; GFX11-NEXT: s_mov_b32 s32, s34
|
|
; GFX11-NEXT: s_mov_b32 s34, s4
|
|
; GFX11-NEXT: s_mov_b32 s33, s3
|
|
; GFX11-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX12-LABEL: s_add_use_fi_sgpr_offset:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-NEXT: s_wait_expcnt 0x0
|
|
; GFX12-NEXT: s_wait_samplecnt 0x0
|
|
; GFX12-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-NEXT: s_wait_kmcnt 0x0
|
|
; GFX12-NEXT: s_mov_b32 s3, s33
|
|
; GFX12-NEXT: s_add_co_i32 s33, s32, 31
|
|
; GFX12-NEXT: s_mov_b32 s4, s34
|
|
; GFX12-NEXT: s_wait_alu depctr_sa_sdst(0)
|
|
; GFX12-NEXT: s_and_not1_b32 s33, s33, 31
|
|
; GFX12-NEXT: s_mov_b32 s34, s32
|
|
; GFX12-NEXT: s_addk_co_i32 s32, 0x1140
|
|
; GFX12-NEXT: s_and_b32 s0, s0, 31
|
|
; GFX12-NEXT: s_wait_alu depctr_sa_sdst(0)
|
|
; GFX12-NEXT: s_add_co_i32 s2, s33, 0x100
|
|
; GFX12-NEXT: s_add_co_i32 s5, s33, 0x100
|
|
; GFX12-NEXT: s_wait_alu depctr_sa_sdst(0)
|
|
; GFX12-NEXT: s_mov_b32 s1, s2
|
|
; GFX12-NEXT: s_mov_b32 s2, s33
|
|
; GFX12-NEXT: s_or_b32 s0, s0, s5
|
|
; GFX12-NEXT: ;;#ASMSTART
|
|
; GFX12-NEXT: ; use s1
|
|
; GFX12-NEXT: ;;#ASMEND
|
|
; GFX12-NEXT: ;;#ASMSTART
|
|
; GFX12-NEXT: ; use s0
|
|
; GFX12-NEXT: ;;#ASMEND
|
|
; GFX12-NEXT: s_mov_b32 s32, s34
|
|
; GFX12-NEXT: s_mov_b32 s34, s4
|
|
; GFX12-NEXT: s_mov_b32 s33, s3
|
|
; GFX12-NEXT: s_wait_alu depctr_sa_sdst(0)
|
|
; GFX12-NEXT: s_setpc_b64 s[30:31]
|
|
%filler = alloca [256 x i8], align 16, addrspace(5)
|
|
%alloca = alloca [4096 x i8], align 32, addrspace(5)
|
|
|
|
%soffset.and = and i32 %soffset, 31
|
|
|
|
call void asm sideeffect "; use $0", "s,s"(ptr addrspace(5) %alloca, ptr addrspace(5) %filler)
|
|
%gep = getelementptr inbounds nuw [4096 x i8], ptr addrspace(5) %alloca, i32 0, i32 %soffset.and
|
|
call void asm sideeffect "; use $0", "s"(ptr addrspace(5) %gep)
|
|
ret void
|
|
}
|