fc0653f31c
Recent upstream trends have moved away from explicitly using `-verify-machineinstrs`, as it's already covered by the expensive checks. This PR removes almost all `-verify-machineinstrs` from tests in `llvm/test/CodeGen/AMDGPU/*.ll`, leaving only those tests where its removal currently causes failures.
50 lines
1.5 KiB
LLVM
50 lines
1.5 KiB
LLVM
; RUN: llc -mtriple=amdgcn < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -mtriple=r600 -mcpu=redwood < %s | FileCheck -check-prefix=R600 -check-prefix=FUNC %s
|
|
|
|
; FUNC-LABEL: {{^}}floor_f32:
|
|
; SI: v_floor_f32_e32
|
|
; R600: FLOOR
|
|
define amdgpu_kernel void @floor_f32(ptr addrspace(1) %out, float %in) {
|
|
%tmp = call float @llvm.floor.f32(float %in) #0
|
|
store float %tmp, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}floor_v2f32:
|
|
; SI: v_floor_f32_e32
|
|
; SI: v_floor_f32_e32
|
|
|
|
define amdgpu_kernel void @floor_v2f32(ptr addrspace(1) %out, <2 x float> %in) {
|
|
%tmp = call <2 x float> @llvm.floor.v2f32(<2 x float> %in) #0
|
|
store <2 x float> %tmp, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}floor_v4f32:
|
|
; SI: v_floor_f32_e32
|
|
; SI: v_floor_f32_e32
|
|
; SI: v_floor_f32_e32
|
|
; SI: v_floor_f32_e32
|
|
|
|
; R600: FLOOR
|
|
; R600: FLOOR
|
|
; R600: FLOOR
|
|
; R600: FLOOR
|
|
define amdgpu_kernel void @floor_v4f32(ptr addrspace(1) %out, <4 x float> %in) {
|
|
%tmp = call <4 x float> @llvm.floor.v4f32(<4 x float> %in) #0
|
|
store <4 x float> %tmp, ptr addrspace(1) %out
|
|
ret void
|
|
}
|
|
|
|
; Function Attrs: nounwind readonly
|
|
declare float @llvm.floor.f32(float) #0
|
|
|
|
; Function Attrs: nounwind readonly
|
|
declare <2 x float> @llvm.floor.v2f32(<2 x float>) #0
|
|
|
|
; Function Attrs: nounwind readonly
|
|
declare <4 x float> @llvm.floor.v4f32(<4 x float>) #0
|
|
|
|
attributes #0 = { nounwind readnone }
|