SPEC CPU result comparison

15th October 2026 - Normalized report

SPECFP 2006: LTO -O2 generic - LLVM and GCC

-O2 -flto

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
410.bwaves 100.00% 101.14%
416.gamess 100.00% 103.93%
433.milc 100.00% 194.86%
434.zeusmp 100.00% 105.19%
435.gromacs 100.00% 118.80%
436.cactusADM 100.00% 114.11%
437.leslie3d 100.00% 107.97%
444.namd 100.00% 100.61%
447.dealII 100.00% 102.17%
450.soplex 100.00% 98.69%
453.povray 100.00% 109.11%
454.calculix 100.00% 103.32%
459.GemsFDTD 100.00% 100.00%
465.tonto 100.00% 99.04%
470.lbm 100.00% 54.12%
481.wrf 100.00% 118.98%
482.sphinx3 RE
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
410.bwaves 100.00% 4.18%
416.gamess 100.00% 100.20%
433.milc 100.00% 81.81%
434.zeusmp 100.00% 124.71%
435.gromacs 100.00% 91.09%
436.cactusADM 100.00% 36.51%
437.leslie3d 100.00% 9.94%
444.namd 100.00% 133.96%
447.dealII 100.00% 94.15%
450.soplex 100.00% 93.85%
453.povray 100.00% 83.39%
454.calculix 100.00% 56.80%
459.GemsFDTD 100.00% 28.18%
465.tonto 100.00% 50.35%
470.lbm 100.00% 81.46%
481.wrf 100.00% 91.22%
482.sphinx3 RE
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM GCC
410.bwaves 100.00% 89.94%
416.gamess 100.00% 105.61%
433.milc 100.00% 197.62%
434.zeusmp 100.00% 106.65%
435.gromacs 100.00% 134.60%
436.cactusADM 100.00% 110.48%
437.leslie3d 100.00% 111.54%
444.namd 100.00% 97.99%
447.dealII 100.00% 99.17%
450.soplex 100.00% 98.09%
453.povray 100.00% 102.97%
454.calculix 100.00% 107.24%
459.GemsFDTD 100.00% 99.20%
465.tonto 100.00% 94.14%
470.lbm 100.00% 43.98%
481.wrf 100.00% 119.32%
482.sphinx3 RE
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
410.bwaves 100.00% 4.19%
416.gamess 100.00% 100.20%
433.milc 100.00% 81.80%
434.zeusmp 100.00% 124.71%
435.gromacs 100.00% 91.09%
436.cactusADM 100.00% 36.51%
437.leslie3d 100.00% 9.95%
444.namd 100.00% 133.96%
447.dealII 100.00% 95.80%
450.soplex 100.00% 93.85%
453.povray 100.00% 83.40%
454.calculix 100.00% 56.80%
459.GemsFDTD 100.00% 28.19%
465.tonto 100.00% 50.35%
470.lbm 100.00% 81.76%
481.wrf 100.00% 91.22%
482.sphinx3 RE
Smaller binary sizes are always better.