[PATCH] D89479: [SimplifyCFG] Be more conservative when speculating in loops. (WIP)

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/test/Transforms/SimplifyCFG/X86/speculate-cttz-ctlz.ll:416
 ; ALL-NEXT:    [[TOBOOL:%.*]] = icmp eq i32 [[X:%.*]], 0
+; ALL-NEXT:    br i1 [[TOBOOL]], label [[COND_END]], label [[COND_TRUE:%.*]]
+; ALL:       cond.true:
----------------
Do you have a better example more like the loops you're seeing performance issues on?

This one looks kind of silly since %x is loop invariant.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D89479/new/

https://reviews.llvm.org/D89479