[Bug target/101935] New: 538.imagick_r LTO -Ofast regression on Zen2 and Kabylake caused by r12-2666-g29f0e955c97
jamborm at gcc dot gnu.org
gcc-bugzilla@gcc.gnu.org
Mon Aug 16 12:42:33 GMT 2021
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101935
Bug ID: 101935
Summary: 538.imagick_r LTO -Ofast regression on Zen2 and
Kabylake caused by r12-2666-g29f0e955c97
Product: gcc
Version: 12.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: jamborm at gcc dot gnu.org
CC: hjl at gcc dot gnu.org
Blocks: 26163
Target Milestone: ---
Host: x86_64-linux
Target: x86_64-linux
When compiled with -flto -Ofast -march=native, the benchmark
538.imagick_r is about 28% slower on AMD zen2 machines (see
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=287.507.0 )
and about 18% slower on Intel Kabylake (see
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=11.507.0 )
I have independently reproduced the regression on a zen2-based desktop
and bisected it down to commit r12-2666-g29f0e955c97 (x86: Update
piecewise move and store).
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)
More information about the Gcc-bugs
mailing list