SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 FPrate: Trunk LTO -Ofast - x86-64-v3 vs native

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

Please note that event though the header of this comparison says -Ofast, benchmark 767.nest_r was additionally compiled with -fno-finite-math-only. Furthermore, benchmarks 737.gmsh_r and 748.flightdm_r were compiled with -fno-fast-math, so a big chunk of -Ofast was in fact disabled. (This might put GCC at a disadvantage compared to LLVM-based compilers because the latter defaults to -fno-math-errno and -fno-trapping-math, in future measurements, we will try to enable these flags.)

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark x86-64-v3 Native
709.cactus_r 100.00% 99.31%
722.palm_r 100.00% 92.82%
731.astcenc_r 100.00% 94.46%
736.ocio_r 100.00% 98.02%
737.gmsh_r 100.00% 100.00%
748.flightdm_r 100.00% 103.01%
749.fotonik3d_r 100.00% 93.87%
765.roms_r 100.00% 90.20%
766.femflow_r 100.00% 81.03%
767.nest_r 100.00% 95.21%
772.marian_r 100.00% 56.25%
782.lbm_r 100.00% 93.64%
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
709.cactus_r 100.00% 99.66%
722.palm_r 100.00% 100.55%
731.astcenc_r 100.00% 89.08%
736.ocio_r 100.00% 101.83%
737.gmsh_r 100.00% 102.23%
748.flightdm_r 100.00% 101.36%
749.fotonik3d_r 100.00% 102.47%
765.roms_r 100.00% 100.10%
766.femflow_r 100.00% 99.32%
767.nest_r 100.00% 101.20%
772.marian_r 100.00% 100.46%
782.lbm_r 100.00% 97.07%
Smaller binary sizes are always better.

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

Run-times

Benchmark x86-64-v3 Native
709.cactus_r 100.00% 101.44%
722.palm_r 100.00% 98.97%
731.astcenc_r 100.00% 101.80%
736.ocio_r 100.00% 99.45%
737.gmsh_r 100.00% 100.00%
748.flightdm_r 100.00% 97.62%
749.fotonik3d_r 100.00% 100.00%
765.roms_r 100.00% 100.36%
766.femflow_r 100.00% 97.40%
767.nest_r 100.00% 99.48%
772.marian_r 100.00% 103.57%
782.lbm_r 100.00% 103.43%
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
709.cactus_r 100.00% 100.56%
722.palm_r 100.00% 100.05%
731.astcenc_r 100.00% 95.69%
736.ocio_r 100.00% 100.85%
737.gmsh_r 100.00% 100.73%
748.flightdm_r 100.00% 100.88%
749.fotonik3d_r 100.00% 101.40%
765.roms_r 100.00% 100.25%
766.femflow_r 100.00% 100.03%
767.nest_r 100.00% 100.73%
772.marian_r 100.00% 100.19%
782.lbm_r 100.00% 101.47%
Smaller binary sizes are always better.