[llvm-bugs] [Bug 36280] [SLPVectorizer] 2-way parallel vectorization hurts perf on x86

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=36280

Sanjay Patel <spatel+llvm at rotateright.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Fixed By Commit(s)|                            |325515
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #10 from Sanjay Patel <spatel+llvm at rotateright.com> ---
To summarize some of the discussion in:
https://reviews.llvm.org/D43079
https://reviews.llvm.org/D42981

The SLP vectorizer is not calculating cost by summing uops, so my analysis in
comment 4 is wrong. Why the SLP vectorizer is calculating cost as it does is
another question. 

https://reviews.llvm.org/rL325515

This commit fixes the problem accidentally, but we have a regression test to
verify that the motivating case for this bug does not regress.

...so resolving as fixed.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20180220/bac9af43/attachment-0001.html>