SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 INTrate: Standard -Ofast native - LLVM and GCC

-Ofast -march=native -mtune=native

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

Run-times

Benchmark LLVM GCC
500.perlbench_r 100.00% 90.65%
502.gcc_r 100.00% 95.11%
505.mcf_r 100.00% 86.72%
520.omnetpp_r 100.00% 97.28%
523.xalancbmk_r 100.00% 89.92%
525.x264_r 100.00% 97.03%
531.deepsjeng_r 100.00% 91.02%
541.leela_r 100.00% 103.16%
548.exchange2_r 100.00% 82.39%
557.xz_r 100.00% 104.29%
LLVM clang version 23.1.3 (https://github.com/llvm/llvm-project.git 0d261d1ca552c95a8f007e061c787ac7132fbcbc), flang version 23.1.3 (https://github.com/llvm/llvm-project.git 0d261d1ca552c95a8f007e061c787ac7132fbcbc)
GCC GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.

Non-debug executable sizes

Benchmark LLVM GCC
500.perlbench_r 100.00% 115.34%
502.gcc_r 100.00% 108.84%
505.mcf_r 100.00% 90.08%
520.omnetpp_r 100.00% 110.75%
523.xalancbmk_r 100.00% 129.04%
525.x264_r 100.00% 106.22%
531.deepsjeng_r 100.00% 99.98%
541.leela_r 100.00% 137.14%
548.exchange2_r 100.00% 22.23%
557.xz_r 100.00% 112.86%
Smaller binary sizes are always better.