[PATCH] D16836: [CodeGenPrepare] Remove load-based heuristic

Zia Ansari via llvm-commits llvm-commits at lists.llvm.org
Wed Feb 24 15:46:36 PST 2016


zansari added a comment.

> I would like to hear from Zia and Mitch before proceeding with this patch because of #7. It would be great to have more perf data. But like I said earlier in the thread, I don't know what our policy is in a situation where a heuristic has lost value for test-suite but still has benefits for other cases.


Just back from a break, so I apologize for getting to the party late. Looks like we see some minor perf drops on our side from this, but I think I'm willing to deal with that through plans of beefing up our heuristics after isel. At a high level, the changes do look like they have potential to introduce cases that may cause regressions, but I don't think that the existing heuristic made much sense to begin with.... So, lgmt..


http://reviews.llvm.org/D16836





More information about the llvm-commits mailing list