SPEC CPU result comparison

15th October 2026 - Normalized report

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

-Ofast -flto -march=native -mtune=native

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% 103.07% 109.43% 100.00% 106.17%
433.milc 100.00% 98.75% 103.33% 100.00% 104.64%
434.zeusmp 100.00% 98.23% 101.77% 100.00% 103.60%
435.gromacs 100.00% 100.00% 102.03% 100.00% 102.03%
436.cactusADM 100.00% 104.92% 107.38% 100.00% 102.34%
437.leslie3d 100.00% 101.30% 100.00% 100.00% 98.72%
444.namd 100.00% 100.00% 101.61% 100.00% 101.61%
447.dealII 100.00% 94.59% 94.59% 100.00% 100.00%
450.soplex 100.00% 100.00% 98.78% 100.00% 98.78%
453.povray RE RE RE RE RE
454.calculix 100.00% 97.00% 99.00% 100.00% 102.06%
459.GemsFDTD 100.00% 100.79% 103.67% 100.00% 102.86%
465.tonto 100.00% 93.42% 93.42% 100.00% 100.00%
470.lbm 100.00% 98.93% 106.42% 100.00% 107.57%
481.wrf 100.00% 93.89% 95.20% 100.00% 101.40%
482.sphinx3 100.00% 99.06% 99.06% 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% 111.80% 111.26% 100.00% 99.51%
416.gamess 100.00% 99.98% 99.93% 100.00% 99.96%
433.milc 100.00% 98.60% 97.20% 100.00% 98.58%
434.zeusmp 100.00% 99.99% 99.99% 100.00% 100.00%
435.gromacs 100.00% 98.66% 97.22% 100.00% 98.55%
436.cactusADM 100.00% 93.66% 93.83% 100.00% 100.18%
437.leslie3d 100.00% 89.44% 89.61% 100.00% 100.19%
444.namd 100.00% 106.99% 104.76% 100.00% 97.91%
447.dealII 100.00% 103.10% 103.92% 100.00% 100.80%
450.soplex 100.00% 104.43% 103.60% 100.00% 99.21%
453.povray RE RE RE RE RE
454.calculix 100.00% 100.11% 98.38% 100.00% 98.27%
459.GemsFDTD 100.00% 98.35% 93.85% 100.00% 95.42%
465.tonto 100.00% 102.92% 95.70% 100.00% 92.99%
470.lbm 100.00% 107.26% 105.02% 100.00% 97.92%
481.wrf 100.00% 94.30% 93.58% 100.00% 99.24%
482.sphinx3 100.00% 106.24% 101.36% 100.00% 95.41%
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% 98.75% 98.75% 100.00% 100.00%
416.gamess 100.00% 99.78% 105.19% 100.00% 105.42%
433.milc 100.00% 101.01% 116.67% 100.00% 115.50%
434.zeusmp 100.00% 98.56% 100.00% 100.00% 101.46%
435.gromacs 100.00% 96.02% 93.78% 100.00% 97.67%
436.cactusADM 100.00% 106.98% 105.12% 100.00% 98.26%
437.leslie3d 100.00% 100.00% 101.31% 100.00% 101.31%
444.namd 100.00% 106.43% 102.22% 100.00% 96.04%
447.dealII 100.00% 91.63% 91.63% 100.00% 100.00%
450.soplex 100.00% 98.19% 98.19% 100.00% 100.00%
453.povray RE RE RE RE RE
454.calculix 100.00% 108.86% 110.76% 100.00% 101.74%
459.GemsFDTD 100.00% 100.65% 100.00% 100.00% 99.35%
465.tonto 100.00% 98.95% 96.86% 100.00% 97.89%
470.lbm 100.00% 102.05% 101.37% 100.00% 99.33%
481.wrf 100.00% 95.81% 94.61% 100.00% 98.75%
482.sphinx3 100.00% 100.00% 98.21% 100.00% 98.21%
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% 119.03% 118.66% 100.00% 99.70%
416.gamess 100.00% 100.07% 100.04% 100.00% 99.97%
433.milc 100.00% 101.44% 98.73% 100.00% 97.33%
434.zeusmp 100.00% 100.00% 100.00% 100.00% 100.00%
435.gromacs 100.00% 101.51% 100.42% 100.00% 98.93%
436.cactusADM 100.00% 104.39% 103.76% 100.00% 99.40%
437.leslie3d 100.00% 101.83% 106.95% 100.00% 105.02%
444.namd 100.00% 109.42% 113.00% 100.00% 103.27%
447.dealII 100.00% 105.85% 106.59% 100.00% 100.70%
450.soplex 100.00% 106.69% 108.30% 100.00% 101.51%
453.povray RE RE RE RE RE
454.calculix 100.00% 104.52% 103.94% 100.00% 99.44%
459.GemsFDTD 100.00% 103.80% 102.31% 100.00% 98.56%
465.tonto 100.00% 104.98% 102.98% 100.00% 98.09%
470.lbm 100.00% 109.63% 105.97% 100.00% 96.66%
481.wrf 100.00% 100.37% 99.44% 100.00% 99.08%
482.sphinx3 100.00% 111.43% 107.47% 100.00% 96.45%
Smaller binary sizes are always better.