https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127243
Bug ID: 127243
Summary: [17 Regression] 9-10% slowdown of milc on Zen2 since
r17-3756-g2c1e97207485e5
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
Blocks: 26163
Target Milestone: ---
As seen here
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=685.70.0
there was a 10% exec time slowdown (113s -> 125s) of the milc SPEC 2006
benchmark when compiled with -O2 -g -flto=128 with PGO on an Zen2
machine. I bisected it to r17-3756-g2c1e97207485e5.
commit 2c1e97207485e54ab6ae085f76c07e62a14440ae
Author: Jovan Dmitrovic <[email protected]>
Date: Sat Aug 29 12:09:51 2026 -0600
[PATCH v4 1/3] ivopts: Remove bias from calculating reg pressure cost.
This is a regression against GCC 14 and 15, but not GCC 16 . See the comparison
here:
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=732.70.0&plot.1=734.70.0&plot.2=689.70.0&plot.3=746.70.0&plot.4=744.70.0&plot.5=762.70.0&plot.6=921.70.0&plot.7=1031.70.0&plot.8=1192.70.0&plot.9=1498.70.0&plot.10=685.70.0&;
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)
