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
562 lines
25 KiB
LLVM
562 lines
25 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx900 -global-isel=0 < %s | FileCheck -check-prefixes=GCN,GFX9,GFX9-SDAG %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx900 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GCN,GFX9,GFX9-GISEL %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=0 < %s | FileCheck -check-prefixes=GCN,GFX10,GFX10-SDAG %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GCN,GFX10,GFX10-GISEL %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 < %s | FileCheck -check-prefixes=GCN,GFX11,GFX11-SDAG %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GCN,GFX11,GFX11-GISEL %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -global-isel=0 < %s | FileCheck -check-prefixes=GCN,GFX12,GFX12-SDAG %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GCN,GFX12,GFX12-GISEL %s
|
|
|
|
define amdgpu_cs void @atomic_load_f32x2_monotonic_agent(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_f32x2_monotonic_agent:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX9-NEXT: global_store_dword v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_f32x2_monotonic_agent:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX10-NEXT: global_store_dword v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-LABEL: atomic_load_f32x2_monotonic_agent:
|
|
; GFX11: ; %bb.0:
|
|
; GFX11-NEXT: flat_load_b64 v[0:1], v[0:1] glc
|
|
; GFX11-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX11-NEXT: global_store_b32 v[2:3], v0, off
|
|
; GFX11-NEXT: s_endpgm
|
|
;
|
|
; GFX12-LABEL: atomic_load_f32x2_monotonic_agent:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: flat_load_b64 v[0:1], v[0:1] scope:SCOPE_DEV
|
|
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX12-NEXT: global_store_b32 v[2:3], v0, off
|
|
; GFX12-NEXT: s_endpgm
|
|
%a0 = load atomic <2 x float>, ptr addrspace(0) %p syncscope("agent") monotonic, align 8
|
|
%num1 = extractelement <2 x float> %a0, i32 0
|
|
%num2 = extractelement <2 x float> %a0, i32 1
|
|
%res = fadd float %num1, %num2
|
|
store float %res, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_f16x2_monotonic_agent(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_f16x2_monotonic_agent:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: flat_load_dword v0, v[0:1] glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_f16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_f16x2_monotonic_agent:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: flat_load_dword v0, v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_add_f16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-SDAG-LABEL: atomic_load_f16x2_monotonic_agent:
|
|
; GFX11-SDAG: ; %bb.0:
|
|
; GFX11-SDAG-NEXT: flat_load_b32 v0, v[0:1] glc
|
|
; GFX11-SDAG-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-SDAG-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX11-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX11-GISEL-LABEL: atomic_load_f16x2_monotonic_agent:
|
|
; GFX11-GISEL: ; %bb.0:
|
|
; GFX11-GISEL-NEXT: flat_load_b32 v0, v[0:1] glc
|
|
; GFX11-GISEL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX11-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-GISEL-NEXT: v_add_f16_e32 v0.l, v0.l, v1.l
|
|
; GFX11-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-GISEL-NEXT: s_endpgm
|
|
;
|
|
; GFX12-SDAG-LABEL: atomic_load_f16x2_monotonic_agent:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: flat_load_b32 v0, v[0:1] scope:SCOPE_DEV
|
|
; GFX12-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-SDAG-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX12-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX12-GISEL-LABEL: atomic_load_f16x2_monotonic_agent:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: flat_load_b32 v0, v[0:1] scope:SCOPE_DEV
|
|
; GFX12-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX12-GISEL-NEXT: v_add_f16_e32 v0.l, v0.l, v1.l
|
|
; GFX12-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-GISEL-NEXT: s_endpgm
|
|
%a0 = load atomic <2 x half>, ptr addrspace(0) %p syncscope("agent") monotonic, align 4
|
|
%num1 = extractelement <2 x half> %a0, i32 0
|
|
%num2 = extractelement <2 x half> %a0, i32 1
|
|
%res = fadd half %num1, %num2
|
|
store half %res, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_i16x2_monotonic_agent(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_i16x2_monotonic_agent:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: flat_load_dword v0, v[0:1] glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_u16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_i16x2_monotonic_agent:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: flat_load_dword v0, v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX10-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-SDAG-LABEL: atomic_load_i16x2_monotonic_agent:
|
|
; GFX11-SDAG: ; %bb.0:
|
|
; GFX11-SDAG-NEXT: flat_load_b32 v0, v[0:1] glc
|
|
; GFX11-SDAG-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX11-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX11-GISEL-LABEL: atomic_load_i16x2_monotonic_agent:
|
|
; GFX11-GISEL: ; %bb.0:
|
|
; GFX11-GISEL-NEXT: flat_load_b32 v0, v[0:1] glc
|
|
; GFX11-GISEL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX11-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v1.l
|
|
; GFX11-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-GISEL-NEXT: s_endpgm
|
|
;
|
|
; GFX12-SDAG-LABEL: atomic_load_i16x2_monotonic_agent:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: flat_load_b32 v0, v[0:1] scope:SCOPE_DEV
|
|
; GFX12-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX12-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX12-GISEL-LABEL: atomic_load_i16x2_monotonic_agent:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: flat_load_b32 v0, v[0:1] scope:SCOPE_DEV
|
|
; GFX12-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX12-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v1.l
|
|
; GFX12-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-GISEL-NEXT: s_endpgm
|
|
%a = load atomic <2 x i16>, ptr addrspace(0) %p syncscope("agent") monotonic, align 4
|
|
%e0 = extractelement <2 x i16> %a, i32 0
|
|
%e1 = extractelement <2 x i16> %a, i32 1
|
|
%sum = add i16 %e0, %e1
|
|
store i16 %sum, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_f16x4_monotonic_agent(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_f16x4_monotonic_agent:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_f16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_mul_f16_sdwa v1, v1, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_add_f16_e32 v0, v0, v1
|
|
; GFX9-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_f16x4_monotonic_agent:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_add_f16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-NEXT: v_mul_f16_sdwa v1, v1, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-NEXT: v_add_f16_e32 v0, v0, v1
|
|
; GFX10-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-SDAG-LABEL: atomic_load_f16x4_monotonic_agent:
|
|
; GFX11-SDAG: ; %bb.0:
|
|
; GFX11-SDAG-NEXT: flat_load_b64 v[0:1], v[0:1] glc
|
|
; GFX11-SDAG-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-SDAG-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX11-SDAG-NEXT: v_mul_f16_e32 v0.h, v1.l, v1.h
|
|
; GFX11-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-SDAG-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX11-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX11-GISEL-LABEL: atomic_load_f16x4_monotonic_agent:
|
|
; GFX11-GISEL: ; %bb.0:
|
|
; GFX11-GISEL-NEXT: flat_load_b64 v[0:1], v[0:1] glc
|
|
; GFX11-GISEL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v4, 16, v0
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX11-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
|
|
; GFX11-GISEL-NEXT: v_add_f16_e32 v0.l, v0.l, v4.l
|
|
; GFX11-GISEL-NEXT: v_mul_f16_e32 v0.h, v1.l, v5.l
|
|
; GFX11-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-GISEL-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX11-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-GISEL-NEXT: s_endpgm
|
|
;
|
|
; GFX12-SDAG-LABEL: atomic_load_f16x4_monotonic_agent:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: flat_load_b64 v[0:1], v[0:1] scope:SCOPE_DEV
|
|
; GFX12-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-SDAG-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX12-SDAG-NEXT: v_mul_f16_e32 v0.h, v1.l, v1.h
|
|
; GFX12-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX12-SDAG-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX12-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX12-GISEL-LABEL: atomic_load_f16x4_monotonic_agent:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: flat_load_b64 v[0:1], v[0:1] scope:SCOPE_DEV
|
|
; GFX12-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v4, 16, v0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
|
|
; GFX12-GISEL-NEXT: v_add_f16_e32 v0.l, v0.l, v4.l
|
|
; GFX12-GISEL-NEXT: v_mul_f16_e32 v0.h, v1.l, v5.l
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX12-GISEL-NEXT: v_add_f16_e32 v0.l, v0.l, v0.h
|
|
; GFX12-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-GISEL-NEXT: s_endpgm
|
|
%a0 = load atomic <4 x half>, ptr addrspace(0) %p syncscope("agent") monotonic, align 8
|
|
%num1 = extractelement <4 x half> %a0, i32 0
|
|
%num2 = extractelement <4 x half> %a0, i32 1
|
|
%num3 = extractelement <4 x half> %a0, i32 2
|
|
%num4 = extractelement <4 x half> %a0, i32 3
|
|
%add = fadd half %num1, %num2
|
|
%mul = fmul half %num3, %num4
|
|
%res = fadd half %add, %mul
|
|
store half %res, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_i16x4_monotonic_agent(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_i16x4_monotonic_agent:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v4, 16, v1
|
|
; GFX9-NEXT: v_add_u16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_mad_legacy_u16 v0, v1, v4, v0
|
|
; GFX9-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_i16x4_monotonic_agent:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_lshrrev_b32_e32 v4, 16, v0
|
|
; GFX10-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX10-NEXT: v_add_nc_u16 v0, v0, v4
|
|
; GFX10-NEXT: v_mad_u16 v0, v1, v5, v0
|
|
; GFX10-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-SDAG-LABEL: atomic_load_i16x4_monotonic_agent:
|
|
; GFX11-SDAG: ; %bb.0:
|
|
; GFX11-SDAG-NEXT: flat_load_b64 v[0:1], v[0:1] glc
|
|
; GFX11-SDAG-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX11-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-SDAG-NEXT: v_mad_u16 v0.l, v1.l, v1.h, v0.l
|
|
; GFX11-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX11-GISEL-LABEL: atomic_load_i16x4_monotonic_agent:
|
|
; GFX11-GISEL: ; %bb.0:
|
|
; GFX11-GISEL-NEXT: flat_load_b64 v[0:1], v[0:1] glc
|
|
; GFX11-GISEL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v4, 16, v0
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX11-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1)
|
|
; GFX11-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v4.l
|
|
; GFX11-GISEL-NEXT: v_mad_u16 v0.l, v1.l, v5.l, v0.l
|
|
; GFX11-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-GISEL-NEXT: s_endpgm
|
|
;
|
|
; GFX12-SDAG-LABEL: atomic_load_i16x4_monotonic_agent:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: flat_load_b64 v[0:1], v[0:1] scope:SCOPE_DEV
|
|
; GFX12-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX12-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX12-SDAG-NEXT: v_mad_u16 v0.l, v1.l, v1.h, v0.l
|
|
; GFX12-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX12-GISEL-LABEL: atomic_load_i16x4_monotonic_agent:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: flat_load_b64 v[0:1], v[0:1] scope:SCOPE_DEV
|
|
; GFX12-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v4, 16, v0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1)
|
|
; GFX12-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v4.l
|
|
; GFX12-GISEL-NEXT: v_mad_u16 v0.l, v1.l, v5.l, v0.l
|
|
; GFX12-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-GISEL-NEXT: s_endpgm
|
|
%a0 = load atomic <4 x i16>, ptr addrspace(0) %p syncscope("agent") monotonic, align 8
|
|
%num1 = extractelement <4 x i16> %a0, i32 0
|
|
%num2 = extractelement <4 x i16> %a0, i32 1
|
|
%num3 = extractelement <4 x i16> %a0, i32 2
|
|
%num4 = extractelement <4 x i16> %a0, i32 3
|
|
%add = add i16 %num1, %num2
|
|
%mul = mul i16 %num3, %num4
|
|
%res = add i16 %add, %mul
|
|
store i16 %res, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_f32x2_monotonic_agent_offset_min(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_f32x2_monotonic_agent_offset_min:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, 0xfffff000, v0
|
|
; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, -1, v1, vcc
|
|
; GFX9-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX9-NEXT: global_store_dword v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_f32x2_monotonic_agent_offset_min:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, 0xfffff000, v0
|
|
; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, -1, v1, vcc_lo
|
|
; GFX10-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX10-NEXT: global_store_dword v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-LABEL: atomic_load_f32x2_monotonic_agent_offset_min:
|
|
; GFX11: ; %bb.0:
|
|
; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, 0xfffff000, v0
|
|
; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-NEXT: v_add_co_ci_u32_e64 v1, null, -1, v1, vcc_lo
|
|
; GFX11-NEXT: flat_load_b64 v[0:1], v[0:1] glc
|
|
; GFX11-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX11-NEXT: global_store_b32 v[2:3], v0, off
|
|
; GFX11-NEXT: s_endpgm
|
|
;
|
|
; GFX12-LABEL: atomic_load_f32x2_monotonic_agent_offset_min:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: flat_load_b64 v[0:1], v[0:1] offset:-4096 scope:SCOPE_DEV
|
|
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX12-NEXT: global_store_b32 v[2:3], v0, off
|
|
; GFX12-NEXT: s_endpgm
|
|
%gep = getelementptr inbounds i8, ptr addrspace(0) %p, i64 -4096
|
|
%a0 = load atomic <2 x float>, ptr addrspace(0) %gep syncscope("agent") monotonic, align 8
|
|
%num1 = extractelement <2 x float> %a0, i32 0
|
|
%num2 = extractelement <2 x float> %a0, i32 1
|
|
%res = fadd float %num1, %num2
|
|
store float %res, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_f32x2_monotonic_agent_offset_max(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_f32x2_monotonic_agent_offset_max:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: flat_load_dwordx2 v[0:1], v[0:1] offset:4095 glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX9-NEXT: global_store_dword v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_f32x2_monotonic_agent_offset_max:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, 0xfff, v0
|
|
; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo
|
|
; GFX10-NEXT: flat_load_dwordx2 v[0:1], v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX10-NEXT: global_store_dword v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-LABEL: atomic_load_f32x2_monotonic_agent_offset_max:
|
|
; GFX11: ; %bb.0:
|
|
; GFX11-NEXT: flat_load_b64 v[0:1], v[0:1] offset:4095 glc
|
|
; GFX11-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX11-NEXT: global_store_b32 v[2:3], v0, off
|
|
; GFX11-NEXT: s_endpgm
|
|
;
|
|
; GFX12-LABEL: atomic_load_f32x2_monotonic_agent_offset_max:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: flat_load_b64 v[0:1], v[0:1] offset:4095 scope:SCOPE_DEV
|
|
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-NEXT: v_add_f32_e32 v0, v0, v1
|
|
; GFX12-NEXT: global_store_b32 v[2:3], v0, off
|
|
; GFX12-NEXT: s_endpgm
|
|
%gep = getelementptr inbounds i8, ptr addrspace(0) %p, i64 4095
|
|
%a0 = load atomic <2 x float>, ptr addrspace(0) %gep syncscope("agent") monotonic, align 8
|
|
%num1 = extractelement <2 x float> %a0, i32 0
|
|
%num2 = extractelement <2 x float> %a0, i32 1
|
|
%res = fadd float %num1, %num2
|
|
store float %res, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_i16x2_monotonic_agent_offset_min(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_i16x2_monotonic_agent_offset_min:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, 0xfffff000, v0
|
|
; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, -1, v1, vcc
|
|
; GFX9-NEXT: flat_load_dword v0, v[0:1] glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_u16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_i16x2_monotonic_agent_offset_min:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, 0xfffff000, v0
|
|
; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, -1, v1, vcc_lo
|
|
; GFX10-NEXT: flat_load_dword v0, v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX10-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-SDAG-LABEL: atomic_load_i16x2_monotonic_agent_offset_min:
|
|
; GFX11-SDAG: ; %bb.0:
|
|
; GFX11-SDAG-NEXT: v_add_co_u32 v0, vcc_lo, 0xfffff000, v0
|
|
; GFX11-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-SDAG-NEXT: v_add_co_ci_u32_e64 v1, null, -1, v1, vcc_lo
|
|
; GFX11-SDAG-NEXT: flat_load_b32 v0, v[0:1] glc
|
|
; GFX11-SDAG-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX11-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX11-GISEL-LABEL: atomic_load_i16x2_monotonic_agent_offset_min:
|
|
; GFX11-GISEL: ; %bb.0:
|
|
; GFX11-GISEL-NEXT: v_add_co_u32 v0, vcc_lo, 0xfffff000, v0
|
|
; GFX11-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1)
|
|
; GFX11-GISEL-NEXT: v_add_co_ci_u32_e64 v1, null, -1, v1, vcc_lo
|
|
; GFX11-GISEL-NEXT: flat_load_b32 v0, v[0:1] glc
|
|
; GFX11-GISEL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX11-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v1.l
|
|
; GFX11-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-GISEL-NEXT: s_endpgm
|
|
;
|
|
; GFX12-SDAG-LABEL: atomic_load_i16x2_monotonic_agent_offset_min:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: flat_load_b32 v0, v[0:1] offset:-4096 scope:SCOPE_DEV
|
|
; GFX12-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX12-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX12-GISEL-LABEL: atomic_load_i16x2_monotonic_agent_offset_min:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: flat_load_b32 v0, v[0:1] offset:-4096 scope:SCOPE_DEV
|
|
; GFX12-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX12-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v1.l
|
|
; GFX12-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-GISEL-NEXT: s_endpgm
|
|
%gep = getelementptr inbounds i8, ptr addrspace(0) %p, i64 -4096
|
|
%a = load atomic <2 x i16>, ptr addrspace(0) %gep syncscope("agent") monotonic, align 8
|
|
%e0 = extractelement <2 x i16> %a, i32 0
|
|
%e1 = extractelement <2 x i16> %a, i32 1
|
|
%sum = add i16 %e0, %e1
|
|
store i16 %sum, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @atomic_load_i16x2_monotonic_agent_offset_max(ptr addrspace(0) %p, ptr addrspace(1) %out) {
|
|
; GFX9-LABEL: atomic_load_i16x2_monotonic_agent_offset_max:
|
|
; GFX9: ; %bb.0:
|
|
; GFX9-NEXT: flat_load_dword v0, v[0:1] offset:4095 glc
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-NEXT: v_add_u16_sdwa v0, v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX10-LABEL: atomic_load_i16x2_monotonic_agent_offset_max:
|
|
; GFX10: ; %bb.0:
|
|
; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, 0xfff, v0
|
|
; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo
|
|
; GFX10-NEXT: flat_load_dword v0, v[0:1] glc dlc
|
|
; GFX10-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX10-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-NEXT: global_store_short v[2:3], v0, off
|
|
; GFX10-NEXT: s_endpgm
|
|
;
|
|
; GFX11-SDAG-LABEL: atomic_load_i16x2_monotonic_agent_offset_max:
|
|
; GFX11-SDAG: ; %bb.0:
|
|
; GFX11-SDAG-NEXT: flat_load_b32 v0, v[0:1] offset:4095 glc
|
|
; GFX11-SDAG-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX11-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX11-GISEL-LABEL: atomic_load_i16x2_monotonic_agent_offset_max:
|
|
; GFX11-GISEL: ; %bb.0:
|
|
; GFX11-GISEL-NEXT: flat_load_b32 v0, v[0:1] offset:4095 glc
|
|
; GFX11-GISEL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX11-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX11-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX11-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v1.l
|
|
; GFX11-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX11-GISEL-NEXT: s_endpgm
|
|
;
|
|
; GFX12-SDAG-LABEL: atomic_load_i16x2_monotonic_agent_offset_max:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: flat_load_b32 v0, v[0:1] offset:4095 scope:SCOPE_DEV
|
|
; GFX12-SDAG-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-SDAG-NEXT: v_add_nc_u16 v0.l, v0.l, v0.h
|
|
; GFX12-SDAG-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-SDAG-NEXT: s_endpgm
|
|
;
|
|
; GFX12-GISEL-LABEL: atomic_load_i16x2_monotonic_agent_offset_max:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: flat_load_b32 v0, v[0:1] offset:4095 scope:SCOPE_DEV
|
|
; GFX12-GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-GISEL-NEXT: v_lshrrev_b32_e32 v1, 16, v0
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
|
|
; GFX12-GISEL-NEXT: v_add_nc_u16 v0.l, v0.l, v1.l
|
|
; GFX12-GISEL-NEXT: global_store_b16 v[2:3], v0, off
|
|
; GFX12-GISEL-NEXT: s_endpgm
|
|
%gep = getelementptr inbounds i8, ptr addrspace(0) %p, i64 4095
|
|
%a = load atomic <2 x i16>, ptr addrspace(0) %gep syncscope("agent") monotonic, align 8
|
|
%e0 = extractelement <2 x i16> %a, i32 0
|
|
%e1 = extractelement <2 x i16> %a, i32 1
|
|
%sum = add i16 %e0, %e1
|
|
store i16 %sum, ptr addrspace(1) %out, align 4
|
|
ret void
|
|
}
|
|
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
|
|
; GCN: {{.*}}
|
|
; GFX10-GISEL: {{.*}}
|
|
; GFX10-SDAG: {{.*}}
|
|
; GFX9-GISEL: {{.*}}
|
|
; GFX9-SDAG: {{.*}}
|