HighW4y2H3ll wrote:

@david-arm Ahhhh, thx for testing! Let me try get some benchmark running on my 
end too. One thing I noticed is that `simplifyLoop` also calls to this 
`foldBranchToCommonDest` function, which is used in many loop passes (e.g. 
InductiveRangeCheckElimination, LoopBoundSplit, LoopFuse, LoopUnroll, LoopPeel, 
LoopVectorize etc.). However, `simplifyLoop` is not passing a 
`TargetTransformInfo` pointer, so neither branch probability nor instruction 
costs will be checked... maybe that could cause regression too...
https://github.com/llvm/llvm-project/blob/2a8c65e983b3f4e1c83d8028d354f7bacc149015/llvm/lib/Transforms/Utils/LoopSimplify.cpp#L659

https://github.com/llvm/llvm-project/pull/144434
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to