[PATCH] D49569: [X86] Remove what appear to be unnecessary uses of DCI.CombineTo

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

I think this might have been to avoid the high cost of calling combineX86ShufflesRecursively again - can you see any compile time diffs? Possibly from one of the vector-shuffle-*.ll test files?


https://reviews.llvm.org/D49569