[PATCH] D54713: [SCEV] Guard movement of insertion point for loop-invariants

[Serguei Katkov via Phabricator via llvm-commits]

skatkov added inline comments.


================
Comment at: llvm/test/Transforms/LoopVectorize/pr30806.ll:8
+;   for(uint32_t outer_i = 0; outer_i < outer_lim; ++outer_i) {
+;     if (denom > 0) {
+;       const uint32_t lim = numer / denom;
----------------
As a side comment, you can consider updating your pipeline with adding a loop unswitch pass somewhere before vectorization. This invariant check should be trivially hoisted from the outer loop.
You'll get better code and will not meet this bug probably :)


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D54713/new/

https://reviews.llvm.org/D54713