SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 FPrate: Trunk LTO native - O3 vs -Ofast

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

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 O3 Ofast
709.cactus_r 100.00% 85.42%
722.palm_r 100.00% 92.08%
731.astcenc_r 100.00% 79.95%
736.ocio_r 100.00% 95.59%
737.gmsh_r 100.00% 100.91%
748.flightdm_r 100.00% 100.00%
749.fotonik3d_r 100.00% 101.32%
765.roms_r 100.00% 90.95%
766.femflow_r 100.00% 100.00%
767.nest_r 100.00% 82.68%
772.marian_r 100.00% 75.00%
782.lbm_r 100.00% 94.93%
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
709.cactus_r 100.00% 97.30%
722.palm_r 100.00% 98.91%
731.astcenc_r 100.00% 122.70%
736.ocio_r 100.00% 100.13%
737.gmsh_r 100.00% 99.99%
748.flightdm_r 100.00% 100.00%
749.fotonik3d_r 100.00% 98.91%
765.roms_r 100.00% 98.78%
766.femflow_r 100.00% 105.29%
767.nest_r 100.00% 100.05%
772.marian_r 100.00% 101.58%
782.lbm_r 100.00% 101.06%
Smaller binary sizes are always better.

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

Run-times

Benchmark O3 Ofast
709.cactus_r 100.00% 80.78%
722.palm_r 100.00% 86.07%
731.astcenc_r 100.00% 79.21%
736.ocio_r 100.00% 97.55%
737.gmsh_r 100.00% 100.00%
748.flightdm_r 100.00% 100.00%
749.fotonik3d_r 100.00% 100.78%
765.roms_r 100.00% 91.69%
766.femflow_r 100.00% 98.68%
767.nest_r 100.00% 83.89%
772.marian_r 100.00% 87.88%
782.lbm_r 100.00% 75.57%
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
709.cactus_r 100.00% 98.46%
722.palm_r 100.00% 101.86%
731.astcenc_r 100.00% 137.41%
736.ocio_r 100.00% 100.76%
737.gmsh_r 100.00% 99.98%
748.flightdm_r 100.00% 100.01%
749.fotonik3d_r 100.00% 98.38%
765.roms_r 100.00% 99.04%
766.femflow_r 100.00% 102.77%
767.nest_r 100.00% 100.02%
772.marian_r 100.00% 101.59%
782.lbm_r 100.00% 104.03%
Smaller binary sizes are always better.