mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-26 04:32:44 +01:00
dd9ab77318
Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
133 lines
4.9 KiB
LLVM
133 lines
4.9 KiB
LLVM
; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -march=r600 -mcpu=cypress < %s | FileCheck -check-prefix=EG -check-prefix=FUNC %s
|
|
|
|
declare float @llvm.ceil.f32(float) nounwind readnone
|
|
declare <2 x float> @llvm.ceil.v2f32(<2 x float>) nounwind readnone
|
|
declare <3 x float> @llvm.ceil.v3f32(<3 x float>) nounwind readnone
|
|
declare <4 x float> @llvm.ceil.v4f32(<4 x float>) nounwind readnone
|
|
declare <8 x float> @llvm.ceil.v8f32(<8 x float>) nounwind readnone
|
|
declare <16 x float> @llvm.ceil.v16f32(<16 x float>) nounwind readnone
|
|
|
|
; FUNC-LABEL: {{^}}fceil_f32:
|
|
; SI: v_ceil_f32_e32
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT:T[0-9]+\.[XYZW]]]
|
|
; EG: CEIL {{\*? *}}[[RESULT]]
|
|
define amdgpu_kernel void @fceil_f32(float addrspace(1)* %out, float %x) {
|
|
%y = call float @llvm.ceil.f32(float %x) nounwind readnone
|
|
store float %y, float addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}fceil_v2f32:
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG: CEIL {{\*? *}}[[RESULT]]
|
|
; EG: CEIL {{\*? *}}[[RESULT]]
|
|
define amdgpu_kernel void @fceil_v2f32(<2 x float> addrspace(1)* %out, <2 x float> %x) {
|
|
%y = call <2 x float> @llvm.ceil.v2f32(<2 x float> %x) nounwind readnone
|
|
store <2 x float> %y, <2 x float> addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}fceil_v3f32:
|
|
; FIXME-SI: v_ceil_f32_e32
|
|
; FIXME-SI: v_ceil_f32_e32
|
|
; FIXME-SI: v_ceil_f32_e32
|
|
; FIXME-EG: v3 is treated as v2 and v1, hence 2 stores
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT1:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT2:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
define amdgpu_kernel void @fceil_v3f32(<3 x float> addrspace(1)* %out, <3 x float> %x) {
|
|
%y = call <3 x float> @llvm.ceil.v3f32(<3 x float> %x) nounwind readnone
|
|
store <3 x float> %y, <3 x float> addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}fceil_v4f32:
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG: CEIL {{\*? *}}[[RESULT]]
|
|
; EG: CEIL {{\*? *}}[[RESULT]]
|
|
; EG: CEIL {{\*? *}}[[RESULT]]
|
|
; EG: CEIL {{\*? *}}[[RESULT]]
|
|
define amdgpu_kernel void @fceil_v4f32(<4 x float> addrspace(1)* %out, <4 x float> %x) {
|
|
%y = call <4 x float> @llvm.ceil.v4f32(<4 x float> %x) nounwind readnone
|
|
store <4 x float> %y, <4 x float> addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}fceil_v8f32:
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT1:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT2:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
define amdgpu_kernel void @fceil_v8f32(<8 x float> addrspace(1)* %out, <8 x float> %x) {
|
|
%y = call <8 x float> @llvm.ceil.v8f32(<8 x float> %x) nounwind readnone
|
|
store <8 x float> %y, <8 x float> addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: {{^}}fceil_v16f32:
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; SI: v_ceil_f32_e32
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT1:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT2:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT3:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG: MEM_RAT_CACHELESS STORE_RAW [[RESULT4:T[0-9]+]]{{\.[XYZW]}}
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT1]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT2]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT3]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT3]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT3]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT3]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT4]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT4]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT4]]
|
|
; EG-DAG: CEIL {{\*? *}}[[RESULT4]]
|
|
define amdgpu_kernel void @fceil_v16f32(<16 x float> addrspace(1)* %out, <16 x float> %x) {
|
|
%y = call <16 x float> @llvm.ceil.v16f32(<16 x float> %x) nounwind readnone
|
|
store <16 x float> %y, <16 x float> addrspace(1)* %out
|
|
ret void
|
|
}
|