SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 FPrate: Trunk LTO -Ofast - x86-64-v3 vs native

-O3 -flto vs -O3 -flto and a given -march

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark x86-64-v3 Native
503.bwaves_r 100.00% 100.00%
507.cactuBSSN_r 100.00% 93.14%
508.namd_r 100.00% 94.09%
510.parest_r 100.00% 97.47%
511.povray_r 100.00% 95.01%
519.lbm_r 100.00% 91.85%
521.wrf_r 100.00% 97.88%
526.blender_r 100.00% 93.96%
527.cam4_r 100.00% 103.16%
538.imagick_r 100.00% 86.24%
544.nab_r 100.00% 100.84%
549.fotonik3d_r 100.00% 95.77%
554.roms_r 100.00% 90.45%
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 x86-64-v3 Native
503.bwaves_r 100.00% 96.88%
507.cactuBSSN_r 100.00% 107.82%
508.namd_r 100.00% 103.62%
510.parest_r 100.00% 103.02%
511.povray_r 100.00% 101.61%
519.lbm_r 100.00% 96.60%
521.wrf_r 100.00% 102.94%
526.blender_r 100.00% 101.68%
527.cam4_r 100.00% 100.24%
538.imagick_r 100.00% 108.23%
544.nab_r 100.00% 101.44%
549.fotonik3d_r 100.00% 102.65%
554.roms_r 100.00% 100.55%
Smaller binary sizes are always better.

AMD EPYC 75F3 (zen3 based high-frequency Milan)

Run-times

Benchmark x86-64-v3 Native
503.bwaves_r 100.00% 100.00%
507.cactuBSSN_r 100.00% 99.19%
508.namd_r 100.00% 100.70%
510.parest_r 100.00% 101.21%
511.povray_r 100.00% 97.36%
519.lbm_r 100.00% 103.62%
521.wrf_r 100.00% 100.00%
526.blender_r 100.00% 100.00%
527.cam4_r 100.00% 100.00%
538.imagick_r 100.00% 100.47%
544.nab_r 100.00% 101.36%
549.fotonik3d_r 100.00% 99.54%
554.roms_r 100.00% 100.72%
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 x86-64-v3 Native
503.bwaves_r 100.00% 100.38%
507.cactuBSSN_r 100.00% 99.14%
508.namd_r 100.00% 100.16%
510.parest_r 100.00% 100.60%
511.povray_r 100.00% 100.28%
519.lbm_r 100.00% 101.47%
521.wrf_r 100.00% 103.56%
526.blender_r 100.00% 100.29%
527.cam4_r 100.00% 100.07%
538.imagick_r 100.00% 100.58%
544.nab_r 100.00% 100.19%
549.fotonik3d_r 100.00% 100.21%
554.roms_r 100.00% 99.82%
Smaller binary sizes are always better.