[PATCH] D32729: LV: Don't vectorize with unknown loop counts on divergent targets

[Matt Arsenault via Phabricator via llvm-commits]

arsenm updated this revision to Diff 109459.
arsenm edited the summary of this revision.
arsenm added a reviewer: hfinkel.
arsenm added a comment.

Base on new OptSize handling. Allow forced vectorization with metadata in case user knows it is dynamically uniform etc.


https://reviews.llvm.org/D32729

Files:
  lib/Transforms/Vectorize/LoopVectorize.cpp
  test/Transforms/LoopVectorize/AMDGPU/divergent-loop-bounds.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D32729.109459.patch
Type: text/x-patch
Size: 14219 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170802/9707eeb4/attachment.bin>