SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 FPrate: LTO -O3 x86-64-v3 - LLVM and GCC

-O3 -flto -march=x86-64-v3

Please note that the same "optimization levels" mean different things to different compilers. For example, LLVM and LLVM-based compiler tend to be much more aggressive at -O2, often with the corresponding consequences to size of the generated binaries and compilation time. GCC is sometimes more aggressive at -O3. LLVM is also often by default less strict when it comes to standard compliance by default and performs optimizations GCC would only do with a special compilation option or at -Ofast. Keep this is mind when comparing different compilers.

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark LLVM GCC
503.bwaves_r 100.00% 93.41%
507.cactuBSSN_r 100.00% 94.50%
508.namd_r 100.00% 102.86%
510.parest_r 100.00% 63.10%
511.povray_r 100.00% 98.01%
519.lbm_r 100.00% 55.87%
521.wrf_r 100.00% 108.65%
526.blender_r 100.00% 99.30%
527.cam4_r 100.00% 89.69%
538.imagick_r 100.00% 69.52%
544.nab_r 100.00% 96.22%
549.fotonik3d_r 100.00% 107.56%
554.roms_r 100.00% 96.44%
LLVM clang version 23.1.3 (/labs/mjambor/git/llvm 0d261d1ca552c95a8f007e061c787ac7132fbcbc), flang version 23.1.3 (/labs/mjambor/git/llvm 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
503.bwaves_r 100.00% 4.72%
507.cactuBSSN_r 100.00% 99.51%
508.namd_r 100.00% 146.02%
510.parest_r 100.00% 94.84%
511.povray_r 100.00% 97.95%
519.lbm_r 100.00% 55.66%
521.wrf_r 100.00% 103.29%
526.blender_r 100.00% 82.80%
527.cam4_r 100.00% 102.54%
538.imagick_r 100.00% 110.98%
544.nab_r 100.00% 103.71%
549.fotonik3d_r 100.00% 37.99%
554.roms_r 100.00% 61.69%
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM GCC
503.bwaves_r 100.00% 97.10%
507.cactuBSSN_r 100.00% 92.74%
508.namd_r 100.00% 99.29%
510.parest_r 100.00% 77.67%
511.povray_r 100.00% 95.97%
519.lbm_r 100.00% 89.11%
521.wrf_r 100.00% 108.99%
526.blender_r 100.00% 101.55%
527.cam4_r 100.00% 87.53%
538.imagick_r 100.00% 85.47%
544.nab_r 100.00% 93.89%
549.fotonik3d_r 100.00% 102.35%
554.roms_r 100.00% 94.64%
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
503.bwaves_r 100.00% 4.73%
507.cactuBSSN_r 100.00% 99.51%
508.namd_r 100.00% 146.02%
510.parest_r 100.00% 95.75%
511.povray_r 100.00% 97.96%
519.lbm_r 100.00% 56.02%
521.wrf_r 100.00% 103.29%
526.blender_r 100.00% 82.81%
527.cam4_r 100.00% 102.54%
538.imagick_r 100.00% 110.98%
544.nab_r 100.00% 103.73%
549.fotonik3d_r 100.00% 38.00%
554.roms_r 100.00% 61.69%
Smaller binary sizes are always better.