SPEC CPU result comparison

15th October 2026 - Normalized report

SPECFP 2006: PGO+LTO generic -O2 - Released GCCs vs Trunk

-O2 -flto -fprofile-use

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
410.bwaves 100.00% 100.00% 100.00% 100.00% 100.00%
416.gamess 100.00% 95.67% 96.49% 100.00% 100.86%
433.milc 100.00% 101.21% 100.00% 100.00% 98.80%
434.zeusmp 100.00% 97.51% 100.00% 100.00% 102.56%
435.gromacs 100.00% 100.83% 100.83% 100.00% 100.00%
436.cactusADM 100.00% 100.72% 102.88% 100.00% 102.14%
437.leslie3d 100.00% 100.00% 100.00% 100.00% 100.00%
444.namd 100.00% 100.00% 100.00% 100.00% 100.00%
447.dealII 100.00% 98.74% 100.00% 100.00% 101.28%
450.soplex 100.00% 98.94% 100.00% 100.00% 101.07%
453.povray 100.00% 97.38% 97.38% 100.00% 100.00%
454.calculix 100.00% 99.16% 104.62% 100.00% 105.51%
459.GemsFDTD 100.00% 99.10% 99.10% 100.00% 100.00%
465.tonto 100.00% 93.94% 95.96% 100.00% 102.15%
470.lbm 100.00% 100.00% 100.00% 100.00% 100.00%
481.wrf 100.00% 100.95% 97.46% 100.00% 96.54%
482.sphinx3 100.00% 100.00% 102.08% 100.00% 102.08%
GCC 15 GCC 15.3.0 (revision 4db0e8df15bef836558857c291c323add11d035c)
GCC 16 GCC 16.2.0 (revision 78d4ac73dd391005b895a6148cd9831e28e1208b)
Trunk GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.
See also continuous gcc branch benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/branch.

Non-debug executable sizes

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
410.bwaves 100.00% 103.36% 102.48% 100.00% 99.15%
416.gamess 100.00% 100.01% 100.01% 100.00% 100.00%
433.milc 100.00% 99.14% 98.60% 100.00% 99.46%
434.zeusmp 100.00% 100.00% 100.00% 100.00% 100.00%
435.gromacs 100.00% 105.34% 103.27% 100.00% 98.04%
436.cactusADM 100.00% 103.07% 102.60% 100.00% 99.55%
437.leslie3d 100.00% 100.20% 99.66% 100.00% 99.46%
444.namd 100.00% 103.58% 103.12% 100.00% 99.56%
447.dealII 100.00% 104.32% 104.28% 100.00% 99.96%
450.soplex 100.00% 103.86% 103.45% 100.00% 99.61%
453.povray 100.00% 103.44% 101.75% 100.00% 98.36%
454.calculix 100.00% 101.78% 101.52% 100.00% 99.75%
459.GemsFDTD 100.00% 102.16% 101.18% 100.00% 99.04%
465.tonto 100.00% 102.63% 102.71% 100.00% 100.08%
470.lbm 100.00% 100.12% 100.97% 100.00% 100.84%
481.wrf 100.00% 103.17% 103.17% 100.00% 100.00%
482.sphinx3 100.00% 101.83% 100.82% 100.00% 99.01%
Smaller binary sizes are always better.

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

Run-times

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
410.bwaves 100.00% 100.00% 100.00% 100.00% 100.00%
416.gamess 100.00% 97.13% 101.84% 100.00% 104.85%
433.milc 100.00% 97.95% 97.95% 100.00% 100.00%
434.zeusmp 100.00% 97.83% 100.00% 100.00% 102.22%
435.gromacs 100.00% 99.38% 100.00% 100.00% 100.63%
436.cactusADM 100.00% 100.82% 98.56% 100.00% 97.75%
437.leslie3d 100.00% 99.16% 99.16% 100.00% 100.00%
444.namd 100.00% 100.00% 100.00% 100.00% 100.00%
447.dealII 100.00% 98.38% 100.00% 100.00% 101.65%
450.soplex 100.00% 98.07% 98.07% 100.00% 100.00%
453.povray 100.00% 96.57% 96.57% 100.00% 100.00%
454.calculix 100.00% 98.32% 104.63% 100.00% 106.42%
459.GemsFDTD 100.00% 100.81% 100.00% 100.00% 99.20%
465.tonto 100.00% 95.00% 95.87% 100.00% 100.92%
470.lbm 100.00% 100.00% 100.00% 100.00% 100.00%
481.wrf 100.00% 98.18% 100.00% 100.00% 101.86%
482.sphinx3 100.00% 97.01% 100.00% 100.00% 103.08%
GCC 15 GCC 15.3.0 (revision 4db0e8df15bef836558857c291c323add11d035c)
GCC 16 GCC 16.2.0 (revision 78d4ac73dd391005b895a6148cd9831e28e1208b)
Trunk GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.
See also continuous gcc branch benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/branch.

Non-debug executable sizes

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
410.bwaves 100.00% 103.35% 102.34% 100.00% 99.02%
416.gamess 100.00% 100.01% 100.01% 100.00% 100.00%
433.milc 100.00% 99.14% 98.55% 100.00% 99.41%
434.zeusmp 100.00% 100.00% 100.00% 100.00% 100.00%
435.gromacs 100.00% 105.34% 103.27% 100.00% 98.04%
436.cactusADM 100.00% 103.06% 102.61% 100.00% 99.57%
437.leslie3d 100.00% 100.20% 99.69% 100.00% 99.49%
444.namd 100.00% 103.60% 103.12% 100.00% 99.53%
447.dealII 100.00% 104.32% 104.26% 100.00% 99.94%
450.soplex 100.00% 103.86% 103.45% 100.00% 99.61%
453.povray 100.00% 103.44% 101.74% 100.00% 98.35%
454.calculix 100.00% 101.78% 101.52% 100.00% 99.75%
459.GemsFDTD 100.00% 102.16% 101.18% 100.00% 99.04%
465.tonto 100.00% 102.63% 102.71% 100.00% 100.08%
470.lbm 100.00% 100.12% 100.96% 100.00% 100.84%
481.wrf 100.00% 103.17% 103.17% 100.00% 100.00%
482.sphinx3 100.00% 101.83% 100.86% 100.00% 99.05%
Smaller binary sizes are always better.