Bug 112915 - [14/15 Regression] 4% exec time regression of 454.calculix on AMD Zen2
Summary: [14/15 Regression] 4% exec time regression of 454.calculix on AMD Zen2
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: target (show other bugs)
Version: 14.0
: P2 normal
Target Milestone: 14.2
Assignee: Not yet assigned to anyone
URL:
Keywords: missed-optimization, needs-bisection
Depends on:
Blocks: spec
  Show dependency treegraph
 
Reported: 2023-12-08 07:38 UTC by Filip Kastl
Modified: 2024-07-11 08:36 UTC (History)
1 user (show)

See Also:
Host: x86_64-linux
Target: x86_64-linux
Build:
Known to work:
Known to fail:
Last reconfirmed:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Filip Kastl 2023-12-08 07:38:21 UTC
Between commits

g:cff1fa6625d1273f

and

g:8b9d0e8cf482287b

there's a slowdown of 454.calculix 2006 SPEC. 11% regression seen on the graph here (Zen2 machine):

https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=288.170.0

I've also replicated this on another Zen2 machine and got 4% slowdown. I haven't seen this slowdown on other architectures.
Comment 1 Filip Kastl 2023-12-08 07:42:41 UTC
I checked again and its actually a 4% regression on both machines, my bad.
Comment 2 Filip Kastl 2023-12-08 07:48:40 UTC
The options are -Ofast -march=native -mtune=native LTO and PGO
Comment 3 Richard Biener 2024-05-07 07:43:08 UTC
GCC 14.1 is being released, retargeting bugs to GCC 14.2.
Comment 4 Andrew Pinski 2024-07-04 02:03:43 UTC
This looks like it was fixed in Jan 2024 around between 02948ced062b730e and 7f7d9c525c694e36  revision (https://lnt.opensuse.org/db_default/v4/SPEC/44304).
Comment 5 Filip Kastl 2024-07-04 13:05:33 UTC
Yes, currently the measurements are back to where they were before this slowdown. They maybe even improved a bit.

I suggest we close this PR. I'll wait a bit in case there are objections and then do it.
Comment 6 Filip Kastl 2024-07-11 08:36:55 UTC
Closing this PR.