SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 FPrate: PGO+LTO native -Ofast - Released GCCs vs Trunk

-Ofast -flto -march=native -mtune=native -fprofile-use

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
503.bwaves_r 100.00% 100.00% 100.00% 100.00% 100.00%
507.cactuBSSN_r 100.00% 100.00% 100.00% 100.00% 100.00%
508.namd_r 100.00% 101.24% 100.00% 100.00% 98.77%
510.parest_r 100.00% 100.85% 100.00% 100.00% 99.16%
511.povray_r 100.00% 100.63% 101.59% 100.00% 100.95%
519.lbm_r 100.00% 98.60% 98.60% 100.00% 100.00%
521.wrf_r 100.00% 98.23% 99.12% 100.00% 100.90%
526.blender_r 100.00% 93.42% 95.39% 100.00% 102.11%
527.cam4_r 100.00% 98.01% 94.89% 100.00% 96.81%
538.imagick_r 100.00% 94.44% 96.88% 100.00% 102.57%
544.nab_r 100.00% 100.74% 86.98% 100.00% 86.34%
549.fotonik3d_r 100.00% 101.07% 100.00% 100.00% 98.94%
554.roms_r 100.00% 100.00% 99.04% 100.00% 99.04%
GCC 15 GCC 15.3.0 (revision 4db0e8df15bef836558857c291c323add11d035c)
GCC 16 GCC 16.2.0 (revision 78d4ac73dd391005b895a6148cd9831e28e1208b)
Trunk GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.
See also continuous gcc branch benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/branch.

Non-debug executable sizes

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
503.bwaves_r 100.00% 108.96% 109.04% 100.00% 100.07%
507.cactuBSSN_r 100.00% 103.65% 98.87% 100.00% 95.39%
508.namd_r 100.00% 100.98% 100.45% 100.00% 99.48%
510.parest_r 100.00% 102.21% 102.45% 100.00% 100.23%
511.povray_r 100.00% 104.23% 101.99% 100.00% 97.85%
519.lbm_r 100.00% 107.45% 108.74% 100.00% 101.19%
521.wrf_r 100.00% 101.20% 100.79% 100.00% 99.59%
526.blender_r 100.00% 100.37% 100.37% 100.00% 100.00%
527.cam4_r 100.00% 100.04% 100.11% 100.00% 100.06%
538.imagick_r 100.00% 103.19% 102.14% 100.00% 98.98%
544.nab_r 100.00% 100.22% 100.28% 100.00% 100.06%
549.fotonik3d_r 100.00% 104.94% 104.61% 100.00% 99.69%
554.roms_r 100.00% 99.77% 95.33% 100.00% 95.54%
Smaller binary sizes are always better.

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

Run-times

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
503.bwaves_r 100.00% 100.00% 100.00% 100.00% 100.00%
507.cactuBSSN_r 100.00% 100.00% 101.13% 100.00% 101.13%
508.namd_r 100.00% 100.46% 98.61% 100.00% 98.16%
510.parest_r 100.00% 98.84% 98.84% 100.00% 100.00%
511.povray_r 100.00% 103.89% 103.33% 100.00% 99.47%
519.lbm_r 100.00% 94.71% 97.49% 100.00% 102.94%
521.wrf_r 100.00% 98.81% 98.81% 100.00% 100.00%
526.blender_r 100.00% 97.29% 103.12% 100.00% 106.00%
527.cam4_r 100.00% 102.94% 98.40% 100.00% 95.58%
538.imagick_r 100.00% 95.02% 101.66% 100.00% 106.99%
544.nab_r 100.00% 100.00% 89.50% 100.00% 89.50%
549.fotonik3d_r 100.00% 100.46% 100.93% 100.00% 100.46%
554.roms_r 100.00% 100.00% 100.00% 100.00% 100.00%
GCC 15 GCC 15.3.0 (revision 4db0e8df15bef836558857c291c323add11d035c)
GCC 16 GCC 16.2.0 (revision 78d4ac73dd391005b895a6148cd9831e28e1208b)
Trunk GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.
See also continuous gcc branch benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/branch.

Non-debug executable sizes

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
503.bwaves_r 100.00% 112.32% 110.90% 100.00% 98.73%
507.cactuBSSN_r 100.00% 103.80% 99.05% 100.00% 95.43%
508.namd_r 100.00% 100.85% 101.03% 100.00% 100.17%
510.parest_r 100.00% 102.33% 102.42% 100.00% 100.09%
511.povray_r 100.00% 104.46% 102.09% 100.00% 97.74%
519.lbm_r 100.00% 108.36% 109.34% 100.00% 100.90%
521.wrf_r 100.00% 101.75% 101.04% 100.00% 99.30%
526.blender_r 100.00% 100.41% 100.44% 100.00% 100.03%
527.cam4_r 100.00% 100.21% 100.25% 100.00% 100.03%
538.imagick_r 100.00% 103.11% 102.22% 100.00% 99.14%
544.nab_r 100.00% 100.39% 100.26% 100.00% 99.87%
549.fotonik3d_r 100.00% 104.91% 104.95% 100.00% 100.03%
554.roms_r 100.00% 102.07% 99.33% 100.00% 97.32%
Smaller binary sizes are always better.