RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 11:13:28 +01:00
Code Issues Projects Releases Wiki Activity

[PowerPC] - No SExt/ZExt needed for count trailing zeros

Browse Source 
This patch corresponds to review:
https://reviews.llvm.org/D25896

It just eliminates the redundant ZExt after a count trailing zeros instruction.

llvm-svn: 285267
This commit is contained in:
Nemanja Ivanovic 2016-10-27 05:17:58 +00:00
parent 531e48b45b
commit 54552bfee6
2 changed files with 58 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
lib/Target/PowerPC/PPCISelDAGToDAG.cpp
View File

@ -4041,8 +4041,9 @@ static bool PeepholePPC64ZExtGather(SDValue Op32,
return true; return true;
} }
// CNTLZW always produces a 64-bit value in [0,32], and so is zero extended. // CNT[LT]ZW always produce a 64-bit value in [0,32], and so is zero extended.
if (Op32.getMachineOpcode() == PPC::CNTLZW) { if (Op32.getMachineOpcode() == PPC::CNTLZW ||
Op32.getMachineOpcode() == PPC::CNTTZW) {
ToPromote.insert(Op32.getNode()); ToPromote.insert(Op32.getNode());
return true; return true;
} }
@ -4237,6 +4238,7 @@ void PPCDAGToDAGISel::PeepholePPC64ZExt() {
case PPC::LHBRX: NewOpcode = PPC::LHBRX8; break; case PPC::LHBRX: NewOpcode = PPC::LHBRX8; break;
case PPC::LWBRX: NewOpcode = PPC::LWBRX8; break; case PPC::LWBRX: NewOpcode = PPC::LWBRX8; break;
case PPC::CNTLZW: NewOpcode = PPC::CNTLZW8; break; case PPC::CNTLZW: NewOpcode = PPC::CNTLZW8; break;
case PPC::CNTTZW: NewOpcode = PPC::CNTTZW8; break;
case PPC::RLWIMI: NewOpcode = PPC::RLWIMI8; break; case PPC::RLWIMI: NewOpcode = PPC::RLWIMI8; break;
case PPC::OR: NewOpcode = PPC::OR8; break; case PPC::OR: NewOpcode = PPC::OR8; break;
case PPC::SELECT_I4: NewOpcode = PPC::SELECT_I8; break; case PPC::SELECT_I4: NewOpcode = PPC::SELECT_I8; break;

54
test/CodeGen/PowerPC/no-ext-with-count-zeros.ll Normal file
View File

@ -0,0 +1,54 @@
; Function Attrs: nounwind readnone
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-unknown \
; RUN: -mcpu=pwr9 < %s | FileCheck %s
define signext i32 @ctw(i32 signext %a) {
entry:
%0 = tail call i32 @llvm.cttz.i32(i32 %a, i1 false)
ret i32 %0
; CHECK-LABEL: ctw
; CHECK: cnttzw 3, 3
; CHECK-NEXT: blr
}
; Function Attrs: nounwind readnone
declare i32 @llvm.cttz.i32(i32, i1)
; Function Attrs: nounwind readnone
define signext i32 @clw(i32 signext %a) {
entry:
%0 = tail call i32 @llvm.ctlz.i32(i32 %a, i1 false)
ret i32 %0
; CHECK-LABEL: clw
; CHECK: cntlzw 3, 3
; CHECK-NEXT: blr
}
; Function Attrs: nounwind readnone
declare i32 @llvm.ctlz.i32(i32, i1)
; Function Attrs: nounwind readnone
define i64 @ctd(i64 %a) {
entry:
%0 = tail call i64 @llvm.cttz.i64(i64 %a, i1 false)
ret i64 %0
; CHECK-LABEL: ctd
; CHECK: cnttzd 3, 3
; CHECK-NEXT: blr
}
; Function Attrs: nounwind readnone
declare i64 @llvm.cttz.i64(i64, i1)
; Function Attrs: nounwind readnone
define i64 @cld(i64 %a) {
entry:
%0 = tail call i64 @llvm.ctlz.i64(i64 %a, i1 false)
ret i64 %0
; CHECK-LABEL: cld
; CHECK: cntlzd 3, 3
; CHECK-NEXT: blr
}
; Function Attrs: nounwind readnone
declare i64 @llvm.ctlz.i64(i64, i1)