[PATCH] D76844: [InstCombine] try to reduce shuffle with bitcasted operand

Eli Friedman via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Mar 26 10:18:15 PDT 2020


efriedma added inline comments.


================
Comment at: llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp:1944
+    if (auto *V = SimplifyShuffleVectorInst(X, UndefValue::get(XType),
+                                            ConstantVector::get(ScaledMaskC),
+                                            XType, SQ.getWithInstruction(&SVI)))
----------------
spatel wrote:
> RKSimon wrote:
> > Its seems a pity to create a speculative ConstantVector like this - can we easily refactor SimplifyShuffleVectorInst to take ArrayRef<int> ?
> The diffs we want are part of D72467.
> @efriedma - is that patch held up on anything? We may not have settled on the scalable vector future, but changing the fixed vector implementation would make things easier for existing code today.
I'll go over it one more time, but I don't think it's held up on anything in particular.  We didn't completely reach consensus on scalable vector shuffles, but the result is almost certainly compatible with that patch.  I'll try to land it today.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D76844/new/

https://reviews.llvm.org/D76844





More information about the llvm-commits mailing list