[LLVMbugs] [Bug 20054] constant vector value splatted at runtime with broadcast instruction instead of loaded (only core-avx2 target?)

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Mon Sep 22 13:58:58 PDT 2014


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

           What    |Removed                     |Added
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #12 from Sanjay Patel <spatel+llvm at rotateright.com> ---

We should generate splats when optimizing for size now. There's a 'TODO' in the
code to see if any non-AVX2 CPUs would be better off generating splats when
optimizing for speed, but it's probably an immeasurable difference, and it will
become less important going forward as fewer AVX-only systems will exist.

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/20140922/30ac671d/attachment.html>

More information about the llvm-bugs mailing list