https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126146
Bug ID: 126146
Summary: [17 Regression] 4% slowdown of exchange2_r on Zen3
since r17-2002-gf2a4e2540221cb
Product: gcc
Version: 17.0
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: rguenth at gcc dot gnu.org
Blocks: 26163
Target Milestone: ---
Host: x86_64-suse-linux
Target: x86_64-suse-linux
Build: x86_64-suse-linux
As seen here
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=694.407.0
there was a 4% exec time slowdown of the exchange2_r SPEC 2017
benchmark when compiled with -O2 -g -flto=128 with PGO on an Zen3
machine. I bisected it to r17-2002-gf2a4e2540221cb.
f2a4e2540221cb731e5b4296885279da6c58576f is the first bad commit
commit f2a4e2540221cb731e5b4296885279da6c58576f
Author: Richard Biener <[email protected]>
Date: Mon Jun 29 13:21:16 2026 +0200
tree-optimization/126008 - redo PR125730 fix
This is a regression against GCC 14, newer versions are slower. See the
comparison
here:
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=707.407.0&plot.1=713.407.0&plot.2=705.407.0&plot.3=700.407.0&plot.4=709.407.0&plot.5=711.407.0&plot.6=740.407.0&plot.7=892.407.0&plot.8=1056.407.0&plot.9=1161.407.0&plot.10=1481.407.0&plot.11=694.407.0&;
There were also these exchange2_r slowdowns in the same timeframe (so probably
caused by the same commit):
4% Zen2 -O2 -g -flto=128 with PGO
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=687.407.0
4% Zen5 -O2 -g -flto=128 with PGO
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=1244.407.0
4% Zen5 -O2 -march=x86-64-v3 -g -flto=128
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=1249.407.0
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2026, 2k17, 2k and 2k6 and
95)
