[llvm-bugs] [Bug 22640] r217953 degraded lsr optimization at -O3 for -march=nehalem in SciMark2's Sparse matmult benchmark

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=22640

Jack Howarth <howarth.mailing.lists at gmail.com> changed:

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

--- Comment #11 from Jack Howarth <howarth.mailing.lists at gmail.com> ---
Clang 3.8 RC1 on Nehalem now produces Sparse matmult results within 2% of the
that from Clang 3.5.1.

clang 3,8 RC1    1183.13+/-0.46

Interestingly, passing "-mllvm -disable-lsr" in addition to "-O3 -march=native"
now heavily regresses the benchmark in Clang 3.8 RC1.

clang 3.8 RC1     579.23+/-7.51
with "-mllvm -disable-lsr"

-- 
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/20160123/c90af11a/attachment.html>