[PATCH] D67805: [LV] Allow vectorization of hot short trip count loops with epilog

[Philip Reames via Phabricator via llvm-commits]

reames added a comment.

Heuristics looks pretty reasonable to me, but I'd really like to have someone more familiar with how all the existing hueristics interact give the actual LGTM.



================
Comment at: llvm/include/llvm/Transforms/Utils/SizeOpts.h:23
 
 /// Returns true if function \p F is suggested to be size-optimized base on the
 /// profile.
----------------
Please update comments to indicate what None value means.  Your verbal explanation was clear, just make sure the code reflects that.

After that, please separate and land the refactoring change involving these two methods.  Then rebase.


================
Comment at: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:7391
   ScalarEpilogueLowering SEL = CM_ScalarEpilogueAllowed;
+  auto IsColdByProfile = llvm::shouldOptimizeForSize(L->getHeader(), PSI, BFI);
   if (Hints.getForce() != LoopVectorizeHints::FK_Enabled &&
----------------
Explicit type this please for readability.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D67805/new/

https://reviews.llvm.org/D67805