https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123204
Bug ID: 123204
Summary: [16 Regression] 5-7% slowdown of xalancbmk{,_r} on
Zen{4,5} since r16-6133-gca8d97f8cb8d28
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Keywords: missed-optimization
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: jmelcr at gcc dot gnu.org
CC: pheeck at gcc dot gnu.org, victorldn at gcc dot gnu.org
Blocks: 26163
Target Milestone: ---
As seen here
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=1288.367.0
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=992.290.0
there was a 5-7% exec time slowdown of the xalancbmk SPEC benchmarks when
compiled with -Ofast -march=native -flto on a Zen4 and Zen5
machine. I bisected it to r16-6133-gca8d97f8cb8d28.
ca8d97f8cb8d282c190183e9c297d5e0490bab4e is the first bad commit
commit ca8d97f8cb8d282c190183e9c297d5e0490bab4e
Author: Victor Do Nascimento <[email protected]>
Date: Thu May 8 15:28:05 2025 +0100
vect: Relax known iteration number constraint
This is a regression against GCC 15. See the comparisons
here:
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=1268.367.0&plot.1=1327.367.0&plot.2=1294.367.0&plot.3=1359.367.0&plot.4=1288.367.0&;
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=940.290.0&plot.1=990.290.0&plot.2=973.290.0&plot.3=1017.290.0&plot.4=1141.290.0&plot.5=992.290.0&;
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2k17, 2k and 2k6 and 95)
