[PATCH] D125398: [ControlHeightReduction] Freeze condition when converting select to branch

Florian Hahn via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed May 11 09:42:16 PDT 2022


fhahn added a subscriber: spatel.
fhahn added inline comments.


================
Comment at: llvm/test/Transforms/PGOProfile/chr.ll:1293
+; CHECK-NEXT:    [[TMP0:%.*]] = or i1 [[V3]], [[V1_NOT]]
+; CHECK-NEXT:    br i1 [[TMP0]], label [[ENTRY_SPLIT_NONCHR:%.*]], label [[BB1:%.*]], !prof [[PROF19:![0-9]+]]
 ; CHECK:       entry.split.nonchr:
----------------
nikic wrote:
> This is the only non-trivial change. Apparently it restores the result to what it was before https://github.com/llvm/llvm-project/commit/1bf8f9e228546bd54ef9739aa808b71b97ea6051, so this is probably fine.
It would be interesting to know what simplification we are missing and why. Looks like @spatel landed 1bf8f9e228546bd54ef9739aa808b71b97ea6051,, perhaps he has an idea?


================
Comment at: llvm/test/Transforms/PGOProfile/chr.ll:1307
+; CHECK-NEXT:    [[TMP1:%.*]] = freeze i1 [[V4]]
+; CHECK-NEXT:    [[TMP2:%.*]] = freeze i1 [[V5]]
+; CHECK-NEXT:    [[DOTDEMORGAN:%.*]] = or i1 [[TMP1]], [[TMP2]]
----------------
Do you think the results would be better if we would freeze J0/I0?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125398/new/

https://reviews.llvm.org/D125398



More information about the llvm-commits mailing list