https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118957
Filip Kastl <pheeck at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Summary|[15 Regression] 5-9% |[15 Regression] 5-9%
|slowdown of 511.povray_r |slowdown of 511.povray_r
|and 453.povray |and 453.povray since
| |r15-7400-gd3ff498c478ace
CC| |hjl at gcc dot gnu.org
--- Comment #4 from Filip Kastl <pheeck at gcc dot gnu.org> ---
I checked on the Zen5 machine and indeed the commit r15-7400-gd3ff498c478ace
introduces a slowdown (~2.3%). (The other machines are currently busy and I
don't want to disturb them.)
So I guess we can be pretty sure that the commit is the reason for these
slowdowns and can Cc H. J. Lu.
Btw I looked through our benchmarking for some povray speedups asociated with
this commit. There are some:
5% speedup
2017 povray_r Zen4 -O2 -march=generic -flto and some hardening options[1]
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=1103.467.0
??% speedup
2006 povray Zen4 -O2 -march=generic -flto and some hardening options[1]
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=1100.160.0
2% speedup
2017 povray_r Zen4 -Ofast -march=native -flto PGO
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=958.467.0
The slowdowns are more numerous and bigger though.
---
[1] -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=3 -fstack-protector-strong
-funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection
