RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-19 11:02:59 +02:00
Code Issues Projects Releases Wiki Activity

[NVPTX] Enable combineRepeatedFPDivisors for NVPTX.

Browse Source 
Reviewers: tra

Subscribers: jholewinski, llvm-commits

Differential Revision: https://reviews.llvm.org/D29477

llvm-svn: 294011
This commit is contained in:
Justin Lebar 2017-02-03 15:13:50 +00:00
parent b6aef04698
commit b818266aed
2 changed files with 46 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
lib/Target/NVPTX/NVPTXISelLowering.h
View File

@ -530,6 +530,8 @@ public:
int &ExtraSteps, bool &UseOneConst,
bool Reciprocal) const override;
unsigned combineRepeatedFPDivisors() const override { return 2; }
bool allowFMA(MachineFunction &MF, CodeGenOpt::Level OptLevel) const;
bool allowUnsafeFPMath(MachineFunction &MF) const;

44
test/CodeGen/NVPTX/fast-math.ll
View File

@ -117,5 +117,49 @@ define float @fcos_approx(float %a) #0 {
ret float %r
}
; CHECK-LABEL: repeated_div_recip_allowed
define float @repeated_div_recip_allowed(i1 %pred, float %a, float %b, float %divisor) {
; CHECK: rcp.rn.f32
; CHECK: mul.rn.f32
; CHECK: mul.rn.f32
%x = fdiv arcp float %a, %divisor
%y = fdiv arcp float %b, %divisor
%z = select i1 %pred, float %x, float %y
ret float %z
}
; CHECK-LABEL: repeated_div_recip_allowed_ftz
define float @repeated_div_recip_allowed_ftz(i1 %pred, float %a, float %b, float %divisor) #1 {
; CHECK: rcp.rn.ftz.f32
; CHECK: mul.rn.ftz.f32
; CHECK: mul.rn.ftz.f32
%x = fdiv arcp float %a, %divisor
%y = fdiv arcp float %b, %divisor
%z = select i1 %pred, float %x, float %y
ret float %z
}
; CHECK-LABEL: repeated_div_fast
define float @repeated_div_fast(i1 %pred, float %a, float %b, float %divisor) #0 {
; CHECK: rcp.approx.f32
; CHECK: mul.f32
; CHECK: mul.f32
%x = fdiv float %a, %divisor
%y = fdiv float %b, %divisor
%z = select i1 %pred, float %x, float %y
ret float %z
}
; CHECK-LABEL: repeated_div_fast_ftz
define float @repeated_div_fast_ftz(i1 %pred, float %a, float %b, float %divisor) #0 #1 {
; CHECK: rcp.approx.ftz.f32
; CHECK: mul.ftz.f32
; CHECK: mul.ftz.f32
%x = fdiv float %a, %divisor
%y = fdiv float %b, %divisor
%z = select i1 %pred, float %x, float %y
ret float %z
}
attributes #0 = { "unsafe-fp-math" = "true" }
attributes #1 = { "nvptx-f32ftz" = "true" }