Summary: | 531.deepsjeng_r is 7% slower at -O2 -march=znver2 than GCC 9 | ||
---|---|---|---|
Product: | gcc | Reporter: | Martin Jambor <jamborm> |
Component: | target | Assignee: | Not yet assigned to anyone <unassigned> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | hubicka, marxin |
Priority: | P3 | ||
Version: | 10.0 | ||
Target Milestone: | --- | ||
Host: | x86_64-linux | Target: | x86_64-linux |
Build: | Known to work: | ||
Known to fail: | Last reconfirmed: | 2020-03-30 00:00:00 | |
Bug Depends on: | |||
Bug Blocks: | 26163 |
Description
Martin Jambor
2020-03-30 10:04:03 UTC
I can confirm on LNT znver2 machine, but the bisection points to a different commit: https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=324.387.0&plot.1=311.387.0&plot.2=348.387.0&plot.3=280.387.0&plot.4=297.387.0& while LNT znver1 machine is not affected and the speed is similar to GCC 9: https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=145.387.0&plot.1=49.387.0&plot.2=79.387.0&plot.3=259.387.0&plot.4=29.387.0& The regression dropped to 1.9% according to my own measurements which also match LNT (linked above). It is peculiar to an unusual option combination, specific to Zen2 (I cannot see it on Zen3 or CascadeLake) and so I think it is unreasonable to expect that anybody will actually want to work on it. And the PR really is mostly fixed, so let me close it as such. |