ba52f06f9d
Update SIMemoryLegalizer and SIInsertWaitcnts to use separate wait instructions per counter (e.g. S_WAIT_LOADCNT) and split VMCNT into separate LOADCNT, SAMPLECNT and BVHCNT counters.
41 lines
2.0 KiB
LLVM
41 lines
2.0 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck -check-prefix=GFX12 %s
|
|
; RUN: llc -global-isel=1 -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck -check-prefix=GFX12 %s
|
|
|
|
define amdgpu_ps float @atomic_min_flt_1d(<8 x i32> inreg %rsrc, float %data, i32 %s) {
|
|
; GFX12-LABEL: atomic_min_flt_1d:
|
|
; GFX12: ; %bb.0: ; %main_body
|
|
; GFX12-NEXT: image_atomic_min_flt v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
|
|
; GFX12-NEXT: s_wait_loadcnt 0x0
|
|
; GFX12-NEXT: ; return to shader part epilog
|
|
main_body:
|
|
%v = call float @llvm.amdgcn.image.atomic.min.flt.1d.f32.f32(float %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
|
|
ret float %v
|
|
}
|
|
|
|
define amdgpu_ps float @atomic_max_flt_1d(<8 x i32> inreg %rsrc, float %data, i32 %s) {
|
|
; GFX12-LABEL: atomic_max_flt_1d:
|
|
; GFX12: ; %bb.0: ; %main_body
|
|
; GFX12-NEXT: image_atomic_max_flt v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
|
|
; GFX12-NEXT: s_wait_loadcnt 0x0
|
|
; GFX12-NEXT: ; return to shader part epilog
|
|
main_body:
|
|
%v = call float @llvm.amdgcn.image.atomic.max.flt.1d.f32.f32(float %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
|
|
ret float %v
|
|
}
|
|
|
|
define amdgpu_ps float @atomic_add_flt_1d(<8 x i32> inreg %rsrc, float %data, i32 %s) {
|
|
; GFX12-LABEL: atomic_add_flt_1d:
|
|
; GFX12: ; %bb.0: ; %main_body
|
|
; GFX12-NEXT: image_atomic_add_flt v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
|
|
; GFX12-NEXT: s_wait_loadcnt 0x0
|
|
; GFX12-NEXT: ; return to shader part epilog
|
|
main_body:
|
|
%v = call float @llvm.amdgcn.image.atomic.add.flt.1d.f32.f32(float %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
|
|
ret float %v
|
|
}
|
|
|
|
declare float @llvm.amdgcn.image.atomic.add.flt.1d.f32.f32(float, i32, <8 x i32>, i32, i32)
|
|
declare float @llvm.amdgcn.image.atomic.min.flt.1d.f32.f32(float, i32, <8 x i32>, i32, i32)
|
|
declare float @llvm.amdgcn.image.atomic.max.flt.1d.f32.f32(float, i32, <8 x i32>, i32, i32)
|