wangtao
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Change type profile for vpermv back to using operand type for the mask argument to match intrinsic behavior. Add a bitcast to the lowering code to convert mask from v8i32 to v8f32 for vpermps. 

llvm-svn: 154798
This commit is contained in:
Craig Topper 2012-04-16 06:43:40 +00:00
parent c0075aa7ff
commit 26d7a94981
2 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -6655,9 +6655,11 @@ X86TargetLowering::LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) const {
for (unsigned i = 0; i != 8; ++i) {
permclMask.push_back(DAG.getConstant((M[i]>=0) ? M[i] : 0, MVT::i32));
}
SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32,
&permclMask[0], 8);
// Bitcast is for VPERMPS since mask is v8i32 but node takes v8f32
return DAG.getNode(X86ISD::VPERMV, dl, VT,
DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32,
&permclMask[0], 8), V1);
DAG.getNode(ISD::BITCAST, dl, VT, Mask), V1);
}
if (V2IsUndef && HasAVX2 && (VT == MVT::v4i64 || VT == MVT::v4f64))

4
llvm/lib/Target/X86/X86InstrFragmentsSIMD.td
View File

@ -124,8 +124,6 @@ def SDTShuff2OpI : SDTypeProfile<1, 2, [SDTCisVec<0>,
SDTCisSameAs<0,1>, SDTCisInt<2>]>;
def SDTShuff3OpI : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
SDTCisSameAs<0,2>, SDTCisInt<3>]>;
def SDTVPermv : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVT<1, v8i32>,
SDTCisSameAs<0,2>]>;
def SDTVBroadcast : SDTypeProfile<1, 1, [SDTCisVec<0>]>;
def SDTBlend : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
@ -157,7 +155,7 @@ def X86Unpckl : SDNode<"X86ISD::UNPCKL", SDTShuff2Op>;
def X86Unpckh : SDNode<"X86ISD::UNPCKH", SDTShuff2Op>;
def X86VPermilp : SDNode<"X86ISD::VPERMILP", SDTShuff2OpI>;
def X86VPermv : SDNode<"X86ISD::VPERMV", SDTVPermv>;
def X86VPermv : SDNode<"X86ISD::VPERMV", SDTShuff2Op>;
def X86VPermi : SDNode<"X86ISD::VPERMI", SDTShuff2OpI>;
def X86VPerm2x128 : SDNode<"X86ISD::VPERM2X128", SDTShuff3OpI>;