SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 INTrate: Trunk LTO -Ofast native - using 128/256/512bit vectors vs default

-Ofast -march=native -mtune=native with and without -mprefer-vector-width=128,256 and/or 512

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark 128 256 Default 256 Default
500.perlbench_r 100.00% 100.66% 100.00% 100.00% 99.34%
502.gcc_r 100.00% 100.00% 100.87% 100.00% 100.87%
505.mcf_r 100.00% 100.56% 101.68% 100.00% 101.11%
520.omnetpp_r 100.00% 101.79% 103.08% 100.00% 101.26%
523.xalancbmk_r 100.00% 96.63% 93.01% 100.00% 96.26%
525.x264_r 100.00% 97.25% 95.69% 100.00% 98.39%
531.deepsjeng_r 100.00% 100.00% 99.43% 100.00% 99.43%
541.leela_r 100.00% 100.00% 98.82% 100.00% 98.82%
548.exchange2_r 100.00% 101.50% 101.50% 100.00% 100.00%
557.xz_r 100.00% 97.95% 97.54% 100.00% 99.58%
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 128 256 Default 256 Default
500.perlbench_r 100.00% 98.17% 98.83% 100.00% 100.67%
502.gcc_r 100.00% 100.98% 101.47% 100.00% 100.49%
505.mcf_r 100.00% 104.18% 109.16% 100.00% 104.78%
520.omnetpp_r 100.00% 99.85% 100.34% 100.00% 100.49%
523.xalancbmk_r 100.00% 99.18% 98.76% 100.00% 99.58%
525.x264_r 100.00% 101.34% 102.15% 100.00% 100.80%
531.deepsjeng_r 100.00% 100.03% 100.11% 100.00% 100.08%
541.leela_r 100.00% 100.08% 100.43% 100.00% 100.34%
548.exchange2_r 100.00% 104.76% 105.89% 100.00% 101.08%
557.xz_r 100.00% 94.00% 94.13% 100.00% 100.13%
Smaller binary sizes are always better.

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

Run-times

Benchmark 128 Default
500.perlbench_r 100.00% 100.91%
502.gcc_r 100.00% 100.00%
505.mcf_r 100.00% 100.00%
520.omnetpp_r 100.00% 99.30%
523.xalancbmk_r 100.00% 98.75%
525.x264_r 100.00% 98.92%
531.deepsjeng_r 100.00% 100.51%
541.leela_r 100.00% 100.47%
548.exchange2_r 100.00% 99.12%
557.xz_r 100.00% 101.69%
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 128 Default
500.perlbench_r 100.00% 98.21%
502.gcc_r 100.00% 101.23%
505.mcf_r 100.00% 105.25%
520.omnetpp_r 100.00% 99.82%
523.xalancbmk_r 100.00% 99.36%
525.x264_r 100.00% 101.48%
531.deepsjeng_r 100.00% 100.03%
541.leela_r 100.00% 99.91%
548.exchange2_r 100.00% 113.74%
557.xz_r 100.00% 94.45%
Smaller binary sizes are always better.