Bug 112549 - [14/15 Regression] 9% exec time regression of 436.cactusADM on Aarch64
Summary: [14/15 Regression] 9% exec time regression of 436.cactusADM on Aarch64
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: middle-end (show other bugs)
Version: 14.0
: P2 normal
Target Milestone: 14.3
Assignee: Not yet assigned to anyone
URL:
Keywords: missed-optimization, needs-bisection
Depends on:
Blocks: spec
  Show dependency tree / graph
 
Reported: 2023-11-15 15:26 UTC by Filip Kastl
Modified: 2024-10-02 09:33 UTC (History)
3 users (show)

See Also:
Host: aarch64
Target: aarch64
Build:
Known to work:
Known to fail:
Last reconfirmed:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Filip Kastl 2023-11-15 15:26:33 UTC 
As seen here:

https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=575.100.0

there's about 9% slowdown between

g:ada871cfadd3f496 (2023-11-04 08:31)

and

g:fd1596f9962569af (2023-11-14 01:27)

when running spec2006 436.cactusADM compiled with -Ofast. The test was run on an Aarch64 machine (-march=armv8.2-a+crypto+fp16+rcpc+dotprod+ssbs).
Comment 1 Andrew Pinski 2023-11-16 07:24:13 UTC 
Do you know what core is this runs on?
Comment 2 Richard Biener 2023-11-16 07:28:26 UTC 
It's a ampere altra IIRC.
Comment 3 Richard Biener 2024-05-07 07:42:43 UTC 
GCC 14.1 is being released, retargeting bugs to GCC 14.2.
Comment 4 Jakub Jelinek 2024-08-01 09:36:15 UTC 
GCC 14.2 is being released, retargeting bugs to GCC 14.3.
Comment 5 Filip Kastl 2024-09-23 08:28:13 UTC 
The slowdown I reported originally disappeared. If you look at the graph, you can see that the benchmark currently runs slower than on g:ada871cfadd3f496 but it's now unclear where the slowdown occured (it was possibly introduced incrementally over multiple commits).

If no one objects, I will close this PR.
Comment 6 Filip Kastl 2024-10-02 09:33:28 UTC 
Marking as RESOLVED FIXED.