Mask, Subtarget, DAG))
return Broadcast;
- // When the shuffle is mirrored between the 128-bit lanes of the unit, we can
- // use lower latency instructions that will operate on both 128-bit lanes.
- SmallVector<int, 2> RepeatedMask;
- if (is128BitLaneRepeatedShuffleMask(MVT::v4i64, Mask, RepeatedMask)) {
- if (V2.isUndef()) {
+ if (V2.isUndef()) {
+ // When the shuffle is mirrored between the 128-bit lanes of the unit, we
+ // can use lower latency instructions that will operate on both lanes.
+ SmallVector<int, 2> RepeatedMask;
+ if (is128BitLaneRepeatedShuffleMask(MVT::v4i64, Mask, RepeatedMask)) {
int PSHUFDMask[] = {-1, -1, -1, -1};
for (int i = 0; i < 2; ++i)
if (RepeatedMask[i] >= 0) {
DAG.getBitcast(MVT::v8i32, V1),
getV4X86ShuffleImm8ForMask(PSHUFDMask, DL, DAG)));
}
- }
- // AVX2 provides a direct instruction for permuting a single input across
- // lanes.
- if (V2.isUndef())
+ // AVX2 provides a direct instruction for permuting a single input across
+ // lanes.
return DAG.getNode(X86ISD::VPERMI, DL, MVT::v4i64, V1,
getV4X86ShuffleImm8ForMask(Mask, DL, DAG));
+ }
// Try to use shift instructions.
if (SDValue Shift = lowerVectorShuffleAsShift(DL, MVT::v4i64, V1, V2, Mask,