SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 FPrate: Trunk LTO native - O3 vs -Ofast

-O3 -flto -march=native vs -Ofast -flto -march=native

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark O3 Ofast
503.bwaves_r 100.00% 90.18%
507.cactuBSSN_r 100.00% 97.73%
508.namd_r 100.00% 99.00%
510.parest_r 100.00% 68.75%
511.povray_r 100.00% 100.00%
519.lbm_r 100.00% 96.40%
521.wrf_r 100.00% 82.50%
526.blender_r 100.00% 97.90%
527.cam4_r 100.00% 95.23%
538.imagick_r 100.00% 60.59%
544.nab_r 100.00% 89.11%
549.fotonik3d_r 100.00% 99.56%
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 options benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/options.

Non-debug executable sizes

Benchmark O3 Ofast
503.bwaves_r 100.00% 93.77%
507.cactuBSSN_r 100.00% 100.15%
508.namd_r 100.00% 113.76%
510.parest_r 100.00% 101.57%
511.povray_r 100.00% 101.70%
519.lbm_r 100.00% 102.84%
521.wrf_r 100.00% 100.41%
526.blender_r 100.00% 100.82%
527.cam4_r 100.00% 100.07%
538.imagick_r 100.00% 116.41%
544.nab_r 100.00% 98.89%
549.fotonik3d_r 100.00% 98.95%
554.roms_r 100.00% 99.22%
Smaller binary sizes are always better.

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

Run-times

Benchmark O3 Ofast
503.bwaves_r 100.00% 79.10%
507.cactuBSSN_r 100.00% 95.82%
508.namd_r 100.00% 100.70%
510.parest_r 100.00% 64.77%
511.povray_r 100.00% 99.77%
519.lbm_r 100.00% 84.74%
521.wrf_r 100.00% 82.13%
526.blender_r 100.00% 105.32%
527.cam4_r 100.00% 93.12%
538.imagick_r 100.00% 74.48%
544.nab_r 100.00% 85.98%
549.fotonik3d_r 100.00% 99.08%
554.roms_r 100.00% 92.67%
The compiler used was GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d).
Naturally, the smaller the run-times, the better.
See also continuous gcc options benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/options.

Non-debug executable sizes

Benchmark O3 Ofast
503.bwaves_r 100.00% 99.20%
507.cactuBSSN_r 100.00% 100.34%
508.namd_r 100.00% 110.42%
510.parest_r 100.00% 101.21%
511.povray_r 100.00% 101.66%
519.lbm_r 100.00% 105.95%
521.wrf_r 100.00% 101.65%
526.blender_r 100.00% 100.63%
527.cam4_r 100.00% 100.20%
538.imagick_r 100.00% 112.40%
544.nab_r 100.00% 98.98%
549.fotonik3d_r 100.00% 98.04%
554.roms_r 100.00% 99.46%
Smaller binary sizes are always better.