[PATCH] D47124: [X86] Remove masking from vpternlog intrinsics. Use a select in IR instead.

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

In https://reviews.llvm.org/D47124#1106705, @craig.topper wrote:

> I don't know that the fast-isel tests would show anything more than the intrinsics tests do since I ran those test cases through the autoupgrade code to get IR with selects. Should I move the intrinsic+select to fast-isel and leave just intrinsic by itself in the *intrinsic.ll test?


If you want to be completely thorough then test -fast-isel with the code from https://reviews.llvm.org/D47125 - the intrinsics tests are more about coverage and checking all intrinsic variants (masks, load folding etc,.).

IMO it's not a critical issue - although teams are oddly sensitive about debug code not using the intrinsics they expect......


Repository:
  rL LLVM

https://reviews.llvm.org/D47124