SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 FPrate: Trunk standard -Ofast - generic vs native

-Ofast vs -Ofast -march=native -mtune=native

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark Generic Native
503.bwaves_r 100.00% 101.39%
507.cactuBSSN_r 100.00% 76.56%
508.namd_r 100.00% 82.74%
510.parest_r 100.00% 91.25%
511.povray_r 100.00% 97.74%
519.lbm_r 100.00% 85.77%
521.wrf_r 100.00% 84.62%
526.blender_r 100.00% 98.60%
527.cam4_r 100.00% 89.27%
538.imagick_r 100.00% 75.43%
544.nab_r 100.00% 94.24%
549.fotonik3d_r 100.00% 98.48%
554.roms_r 100.00% 68.60%
The compiler used was GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d).
Naturally, the smaller the run-times, the better.
See also continuous gcc tuning benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/tuning.

Non-debug executable sizes

Benchmark Generic Native
503.bwaves_r 100.00% 101.30%
507.cactuBSSN_r 100.00% 101.19%
508.namd_r 100.00% 115.26%
510.parest_r 100.00% 130.32%
511.povray_r 100.00% 103.17%
519.lbm_r 100.00% 90.36%
521.wrf_r 100.00% 107.77%
526.blender_r 100.00% 103.01%
527.cam4_r 100.00% 100.53%
538.imagick_r 100.00% 124.06%
544.nab_r 100.00% 100.88%
549.fotonik3d_r 100.00% 120.34%
554.roms_r 100.00% 106.52%
Smaller binary sizes are always better.