[PATCH] D27573: [X86] Fix bug in r288804.

[Michael Kuperstein via Phabricator via llvm-commits]

mkuper added a comment.

A drive-by comment - regardless of whether we should be forming a vbroadcast for this case (I'll let you and Zvi figure it out :-) ), shouldn't we also have a pattern in place for the case we do?

X86InstrAVX512.td has a lot of X86SubVBroadcast patterns that are commented with:
// Provide fallback in case the load node that is used in the patterns above
// is used by additional users, which prevents the pattern selection.
But I don't see such a pattern for the v4i64 -> v8i64 case.


https://reviews.llvm.org/D27573