53 lines
2.3 KiB
LLVM
53 lines
2.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -global-isel=0 -mtriple=amdgcn -mcpu=gfx1200 -verify-machineinstrs | FileCheck %s -check-prefixes=GFX12,GFX12-SDAG
|
|
; RUN: llc < %s -global-isel=1 -mtriple=amdgcn -mcpu=gfx1200 -verify-machineinstrs | FileCheck %s -check-prefixes=GFX12,GFX12-GISEL
|
|
|
|
declare float @llvm.amdgcn.flat.atomic.fmin.num.f32.p1.f32(ptr %ptr, float %data)
|
|
declare float @llvm.amdgcn.flat.atomic.fmax.num.f32.p1.f32(ptr %ptr, float %data)
|
|
|
|
define amdgpu_cs void @flat_atomic_fmin_num_f32_noret(ptr %ptr, float %data) {
|
|
; GFX12-LABEL: flat_atomic_fmin_num_f32_noret:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: flat_atomic_min_num_f32 v[0:1], v2
|
|
; GFX12-NEXT: s_endpgm
|
|
%ret = call float @llvm.amdgcn.flat.atomic.fmin.num.f32.p1.f32(ptr %ptr, float %data)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs void @flat_atomic_fmax_num_f32_noret(ptr %ptr, float %data) {
|
|
; GFX12-LABEL: flat_atomic_fmax_num_f32_noret:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: flat_atomic_max_num_f32 v[0:1], v2
|
|
; GFX12-NEXT: s_endpgm
|
|
%ret = call float @llvm.amdgcn.flat.atomic.fmax.num.f32.p1.f32(ptr %ptr, float %data)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_cs float @flat_atomic_fmin_num_f32_rtn(ptr %ptr, float %data, ptr %out) {
|
|
; GFX12-LABEL: flat_atomic_fmin_num_f32_rtn:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: flat_atomic_min_num_f32 v0, v[0:1], v2 th:TH_ATOMIC_RETURN
|
|
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-NEXT: flat_store_b32 v[3:4], v0
|
|
; GFX12-NEXT: s_wait_dscnt 0x0
|
|
; GFX12-NEXT: ; return to shader part epilog
|
|
%ret = call float @llvm.amdgcn.flat.atomic.fmin.num.f32.p1.f32(ptr %ptr, float %data)
|
|
store float %ret, ptr %out
|
|
ret float %ret
|
|
}
|
|
|
|
define amdgpu_cs float @flat_atomic_fmax_num_f32_rtn(ptr %ptr, float %data, ptr %out) {
|
|
; GFX12-LABEL: flat_atomic_fmax_num_f32_rtn:
|
|
; GFX12: ; %bb.0:
|
|
; GFX12-NEXT: flat_atomic_max_num_f32 v0, v[0:1], v2 th:TH_ATOMIC_RETURN
|
|
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
|
|
; GFX12-NEXT: flat_store_b32 v[3:4], v0
|
|
; GFX12-NEXT: s_wait_dscnt 0x0
|
|
; GFX12-NEXT: ; return to shader part epilog
|
|
%ret = call float @llvm.amdgcn.flat.atomic.fmax.num.f32.p1.f32(ptr %ptr, float %data)
|
|
store float %ret, ptr %out
|
|
ret float %ret
|
|
}
|
|
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
|
|
; GFX12-GISEL: {{.*}}
|
|
; GFX12-SDAG: {{.*}}
|