mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-12-16 11:30:51 +00:00
de929f8b7d
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213844 91177308-0d34-0410-b5e6-96231b3b80d8
48 lines
1.9 KiB
LLVM
48 lines
1.9 KiB
LLVM
; RUN: llc -march=r600 -mcpu=SI -mattr=-fp32-denormals -enable-unsafe-fp-math -verify-machineinstrs < %s | FileCheck -check-prefix=SI-UNSAFE -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -march=r600 -mcpu=SI -mattr=-fp32-denormals -verify-machineinstrs < %s | FileCheck -check-prefix=SI-SAFE -check-prefix=SI -check-prefix=FUNC %s
|
|
; XUN: llc -march=r600 -mcpu=SI -mattr=+fp32-denormals -verify-machineinstrs < %s | FileCheck -check-prefix=SI-SAFE-SPDENORM -check-prefix=SI -check-prefix=FUNC %s
|
|
|
|
; RUN: llc -march=r600 -mcpu=cypress -verify-machineinstrs < %s | FileCheck -check-prefix=EG-SAFE -check-prefix=FUNC %s
|
|
; RUN: llc -march=r600 -mcpu=cayman -verify-machineinstrs < %s | FileCheck -check-prefix=EG -check-prefix=FUNC %s
|
|
|
|
declare float @llvm.AMDGPU.rcp.f32(float) nounwind readnone
|
|
declare double @llvm.AMDGPU.rcp.f64(double) nounwind readnone
|
|
|
|
declare float @llvm.sqrt.f32(float) nounwind readnone
|
|
|
|
; FUNC-LABEL: @rcp_f32
|
|
; SI: V_RCP_F32_e32
|
|
; EG: RECIP_IEEE
|
|
define void @rcp_f32(float addrspace(1)* %out, float %src) nounwind {
|
|
%rcp = call float @llvm.AMDGPU.rcp.f32(float %src) nounwind readnone
|
|
store float %rcp, float addrspace(1)* %out, align 4
|
|
ret void
|
|
}
|
|
|
|
; FIXME: Evergreen only ever does unsafe fp math.
|
|
; FUNC-LABEL: @rcp_pat_f32
|
|
|
|
; SI-SAFE: V_RCP_F32_e32
|
|
; XSI-SAFE-SPDENORM-NOT: V_RCP_F32_e32
|
|
|
|
; EG: RECIP_IEEE
|
|
|
|
define void @rcp_pat_f32(float addrspace(1)* %out, float %src) nounwind {
|
|
%rcp = fdiv float 1.0, %src
|
|
store float %rcp, float addrspace(1)* %out, align 4
|
|
ret void
|
|
}
|
|
|
|
; FUNC-LABEL: @rsq_rcp_pat_f32
|
|
; SI-UNSAFE: V_RSQ_F32_e32
|
|
; SI-SAFE: V_SQRT_F32_e32
|
|
; SI-SAFE: V_RCP_F32_e32
|
|
|
|
; EG: RECIPSQRT_IEEE
|
|
define void @rsq_rcp_pat_f32(float addrspace(1)* %out, float %src) nounwind {
|
|
%sqrt = call float @llvm.sqrt.f32(float %src) nounwind readnone
|
|
%rcp = call float @llvm.AMDGPU.rcp.f32(float %sqrt) nounwind readnone
|
|
store float %rcp, float addrspace(1)* %out, align 4
|
|
ret void
|
|
}
|