[PATCH] D16435: [RS4GC] Effective rematerialization at non-entry polls

[Philip Reames via llvm-commits]

reames added a comment.

I think I've found a better fix for this issue.  I'm working on a patch for CGP which lets it cleanup the phi cycles introduced by the simple rematerialization.  Assuming that works out, I'll abandon this patch.


http://reviews.llvm.org/D16435