https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125930
Bug ID: 125930
Summary: [17 Regression] 4-6% slowdown of fotonik3d_r on Zen3
since r17-1586-gc36539e6fd0dfa
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=477.527.0
there was a 4-6% exec time slowdown of the fotonik3d_r SPEC 2017
benchmark when compiled with -Ofast -march=native -g -flto=128 with PGO on an
Zen3
machine. I bisected it to r17-1586-gc36539e6fd0dfa.
c36539e6fd0dfa6bd1ebb891df68da9a5bd96163 is the first bad commit
commit c36539e6fd0dfa6bd1ebb891df68da9a5bd96163
Author: Richard Biener <[email protected]>
Date: Thu Jun 11 15:51:02 2026 +0200
tree-optimization/125730 - avoid losing track of base in IVOPTs
This is a regression against GCC 16 (short light blue line in between the trunk
and 15 line). See the comparison
here:
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=550.527.0&plot.1=537.527.0&plot.2=560.527.0&plot.3=481.527.0&plot.4=518.527.0&plot.5=500.527.0&plot.6=862.527.0&plot.7=891.527.0&plot.8=1053.527.0&plot.9=1165.527.0&plot.10=1504.527.0&plot.11=477.527.0&;
There were also these fotonik3d_r slowdowns in the same timeframe (so probably
caused by the same commit):
4% Zen4 -O2 -march=x86-64-v3 -g -flto=128 PGO
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=1108.527.0
5% Zen4 -Ofast -march=native -g -flto=128 (without PGO)
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=959.527.0
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)
