RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2024-12-16 00:16:50 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Allow vector fsub ops with constants to get the same optimizations as scalars.

Browse Source 
This problem is bigger than just fsub, but this is the minimum fix to solve
fneg for PR20556 ( http://llvm.org/bugs/show_bug.cgi?id=20556 ), and we solve
zero subtraction with the same change.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@217286 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Sanjay Patel 2014-09-05 22:26:22 +00:00
parent e8b19acded
commit 52af82df95
2 changed files with 25 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
lib/CodeGen/SelectionDAG/DAGCombiner.cpp
View File

@ -6754,8 +6754,8 @@ SDValue DAGCombiner::visitFADD(SDNode *N) {
SDValue DAGCombiner::visitFSUB(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
ConstantFPSDNode *N1CFP = dyn_cast<ConstantFPSDNode>(N1);
ConstantFPSDNode *N0CFP = isConstOrConstSplatFP(N0);
ConstantFPSDNode *N1CFP = isConstOrConstSplatFP(N1);
EVT VT = N->getValueType(0);
SDLoc dl(N);
const TargetOptions &Options = DAG.getTarget().Options;

23
test/CodeGen/X86/vec_unsafe-fp-math.ll Normal file
View File

@ -0,0 +1,23 @@
; RUN: llc < %s -enable-unsafe-fp-math -mtriple=x86_64-unknown-unknown -mcpu=corei7 | FileCheck %s
; Make sure that vectors get the same benefits as scalars when using unsafe-fp-math.
; Subtracting zero is free.
define <4 x float> @vec_fsub_zero(<4 x float> %x) {
; CHECK-LABEL: vec_fsub_zero:
; CHECK-NOT: subps
; CHECK-NOT: xorps
; CHECK: retq
%sub = fsub <4 x float> %x, zeroinitializer
ret <4 x float> %sub
}
; Negating doesn't require subtraction.
define <4 x float> @vec_fneg(<4 x float> %x) {
; CHECK-LABEL: vec_fneg:
; CHECK: xorps {{.*}}LCP{{.*}}, %xmm0
; CHECK-NOT: subps
; CHECK-NEXT: retq
%sub = fsub <4 x float> zeroinitializer, %x
ret <4 x float> %sub
}