SPEC CPU result comparison

15th October 2026 - Normalized report

SPECFP 2006: PGO+LTO native -Ofast - Released GCCs vs Trunk

-Ofast -flto -march=native -mtune=native -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% 101.71% 102.14% 100.00% 100.42%
433.milc 100.00% 98.78% 102.45% 100.00% 103.72%
434.zeusmp 100.00% 100.00% 101.46% 100.00% 101.46%
435.gromacs 100.00% 98.98% 101.02% 100.00% 102.07%
436.cactusADM 100.00% 110.08% 110.08% 100.00% 100.00%
437.leslie3d 100.00% 101.58% 101.58% 100.00% 100.00%
444.namd 100.00% 100.00% 101.37% 100.00% 101.37%
447.dealII 100.00% 98.90% 101.10% 100.00% 102.22%
450.soplex 100.00% 99.01% 96.79% 100.00% 97.76%
453.povray FE FE FE FE FE
454.calculix 100.00% 103.27% 98.13% 100.00% 95.02%
459.GemsFDTD 100.00% 100.00% 100.00% 100.00% 100.00%
465.tonto 100.00% 91.77% 91.77% 100.00% 100.00%
470.lbm 100.00% 100.00% 121.11% 100.00% 121.11%
481.wrf 100.00% 95.63% 94.05% 100.00% 98.34%
482.sphinx3 100.00% 97.10% 97.10% 100.00% 100.00%
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% 110.84% 110.14% 100.00% 99.37%
416.gamess 100.00% 100.01% 100.01% 100.00% 100.00%
433.milc 100.00% 101.84% 101.71% 100.00% 99.87%
434.zeusmp 100.00% 100.00% 100.00% 100.00% 100.00%
435.gromacs 100.00% 110.26% 104.35% 100.00% 94.64%
436.cactusADM 100.00% 99.84% 99.41% 100.00% 99.57%
437.leslie3d 100.00% 90.48% 88.57% 100.00% 97.88%
444.namd 100.00% 102.48% 103.46% 100.00% 100.96%
447.dealII 100.00% 105.67% 105.45% 100.00% 99.79%
450.soplex 100.00% 104.10% 103.99% 100.00% 99.89%
453.povray FE FE FE FE FE
454.calculix 100.00% 100.97% 100.67% 100.00% 99.70%
459.GemsFDTD 100.00% 101.20% 99.05% 100.00% 97.88%
465.tonto 100.00% 104.66% 102.76% 100.00% 98.18%
470.lbm 100.00% 107.84% 109.34% 100.00% 101.39%
481.wrf 100.00% 101.77% 101.22% 100.00% 99.46%
482.sphinx3 100.00% 101.23% 100.22% 100.00% 99.00%
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% 98.77% 98.77% 100.00% 100.00%
416.gamess 100.00% 97.75% 97.13% 100.00% 99.37%
433.milc 100.00% 104.04% 129.29% 100.00% 124.27%
434.zeusmp 100.00% 95.69% 98.56% 100.00% 103.00%
435.gromacs 100.00% 97.21% 96.45% 100.00% 99.22%
436.cactusADM 100.00% 98.21% 103.14% 100.00% 105.02%
437.leslie3d 100.00% 98.97% 98.97% 100.00% 100.00%
444.namd 100.00% 104.46% 99.55% 100.00% 95.30%
447.dealII 100.00% 98.25% 98.25% 100.00% 100.00%
450.soplex 100.00% 100.00% 98.97% 100.00% 98.97%
453.povray FE FE FE FE FE
454.calculix 100.00% 98.72% 100.00% 100.00% 101.30%
459.GemsFDTD 100.00% 99.36% 100.00% 100.00% 100.65%
465.tonto 100.00% 98.56% 95.32% 100.00% 96.72%
470.lbm 100.00% 100.00% 101.50% 100.00% 101.50%
481.wrf 100.00% 96.48% 95.12% 100.00% 98.60%
482.sphinx3 100.00% 98.94% 101.33% 100.00% 102.41%
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% 113.98% 113.50% 100.00% 99.59%
416.gamess 100.00% 100.01% 100.01% 100.00% 100.00%
433.milc 100.00% 102.82% 102.59% 100.00% 99.77%
434.zeusmp 100.00% 100.00% 100.00% 100.00% 100.00%
435.gromacs 100.00% 109.91% 104.32% 100.00% 94.92%
436.cactusADM 100.00% 102.99% 102.22% 100.00% 99.25%
437.leslie3d 100.00% 102.25% 103.31% 100.00% 101.03%
444.namd 100.00% 103.84% 106.50% 100.00% 102.57%
447.dealII 100.00% 106.36% 106.33% 100.00% 99.97%
450.soplex 100.00% 104.15% 103.54% 100.00% 99.41%
453.povray FE FE FE FE FE
454.calculix 100.00% 101.28% 101.40% 100.00% 100.12%
459.GemsFDTD 100.00% 102.99% 101.27% 100.00% 98.33%
465.tonto 100.00% 104.26% 103.19% 100.00% 98.97%
470.lbm 100.00% 110.23% 111.85% 100.00% 101.47%
481.wrf 100.00% 102.91% 101.84% 100.00% 98.96%
482.sphinx3 100.00% 103.59% 102.80% 100.00% 99.24%
Smaller binary sizes are always better.