mirror of
https://github.com/RPCSX/llvm.git
synced 2025-01-10 06:00:30 +00:00
6ab99c7ca6
Summary: This includes a hazard recognizer implementation to replace some of the hazard handling we had during frame index elimination. Reviewers: arsenm Subscribers: qcolombet, arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D18602 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@268143 91177308-0d34-0410-b5e6-96231b3b80d8
44 lines
1.7 KiB
LLVM
44 lines
1.7 KiB
LLVM
; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=VI -check-prefix=FUNC %s
|
|
|
|
declare float @llvm.amdgcn.rsq.clamp.f32(float) #1
|
|
declare double @llvm.amdgcn.rsq.clamp.f64(double) #1
|
|
|
|
; FUNC-LABEL: {{^}}rsq_clamp_f32:
|
|
; SI: v_rsq_clamp_f32_e32
|
|
|
|
; VI: s_load_dword [[SRC:s[0-9]+]]
|
|
; VI-DAG: v_rsq_f32_e32 [[RSQ:v[0-9]+]], [[SRC]]
|
|
; VI-DAG: v_min_f32_e32 [[MIN:v[0-9]+]], 0x7f7fffff, [[RSQ]]
|
|
; TODO: this constant should be folded:
|
|
; VI-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 0xff7fffff
|
|
; VI: v_max_f32_e32 [[RESULT:v[0-9]+]], [[MIN]], [[K]]
|
|
; VI: buffer_store_dword [[RESULT]]
|
|
define void @rsq_clamp_f32(float addrspace(1)* %out, float %src) #0 {
|
|
%rsq_clamp = call float @llvm.amdgcn.rsq.clamp.f32(float %src)
|
|
store float %rsq_clamp, float addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
|
|
; FUNC-LABEL: {{^}}rsq_clamp_f64:
|
|
; SI: v_rsq_clamp_f64_e32
|
|
|
|
; TODO: this constant should be folded:
|
|
; VI-DAG: s_mov_b32 s[[ALLBITS:[0-9+]]], -1
|
|
; VI-DAG: s_mov_b32 s[[HIGH1:[0-9+]]], 0x7fefffff
|
|
; VI-DAG: s_mov_b32 s[[HIGH2:[0-9+]]], 0xffefffff
|
|
; VI-DAG: s_mov_b32 s[[LOW1:[0-9+]]], s[[ALLBITS]]
|
|
; VI-DAG: v_rsq_f64_e32 [[RSQ:v\[[0-9]+:[0-9]+\]]], s[{{[0-9]+:[0-9]+}}
|
|
; VI: s_mov_b32 s[[LOW2:[0-9+]]], s[[ALLBITS]]
|
|
; VI: v_min_f64 v[0:1], [[RSQ]], s{{\[}}[[LOW1]]:[[HIGH1]]]
|
|
; VI: v_max_f64 v[0:1], v[0:1], s{{\[}}[[LOW2]]:[[HIGH2]]]
|
|
define void @rsq_clamp_f64(double addrspace(1)* %out, double %src) #0 {
|
|
%rsq_clamp = call double @llvm.amdgcn.rsq.clamp.f64(double %src)
|
|
store double %rsq_clamp, double addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
attributes #0 = { nounwind }
|
|
attributes #1 = { nounwind readnone }
|