SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 FPrate: 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
709.cactus_r 100.00% 118.96% 115.90% 100.00% 97.43%
722.palm_r 100.00% 100.00% 98.43% 100.00% 98.43%
731.astcenc_r 100.00% 100.85% 99.36% 100.00% 98.53%
736.ocio_r 100.00% 100.65% 100.00% 100.00% 99.35%
737.gmsh_r 100.00% 101.80% 101.80% 100.00% 100.00%
748.flightdm_r 100.00% 98.96% 98.96% 100.00% 100.00%
749.fotonik3d_r 100.00% 101.57% 100.00% 100.00% 98.46%
765.roms_r 100.00% 100.00% 99.42% 100.00% 99.42%
766.femflow_r 100.00% 81.01% 79.75% 100.00% 98.44%
767.nest_r 100.00% 119.49% 122.05% 100.00% 102.15%
772.marian_r 100.00% 92.97% 91.89% 100.00% 98.84%
782.lbm_r 100.00% 101.31% 101.31% 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
709.cactus_r 100.00% 102.61% 103.56% 100.00% 100.92%
722.palm_r 100.00% 102.86% 102.45% 100.00% 99.60%
731.astcenc_r 100.00% 99.03% 96.83% 100.00% 97.78%
736.ocio_r 100.00% 100.58% 100.66% 100.00% 100.08%
737.gmsh_r 100.00% 101.52% 100.96% 100.00% 99.45%
748.flightdm_r 100.00% 102.24% 101.15% 100.00% 98.93%
749.fotonik3d_r 100.00% 105.28% 104.59% 100.00% 99.34%
765.roms_r 100.00% 101.99% 100.85% 100.00% 98.89%
766.femflow_r 100.00% 102.23% 102.51% 100.00% 100.27%
767.nest_r 100.00% 100.65% 101.08% 100.00% 100.42%
772.marian_r 100.00% 100.61% 100.90% 100.00% 100.29%
782.lbm_r 100.00% 100.00% 100.51% 100.00% 100.51%
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
709.cactus_r 100.00% 111.41% 114.39% 100.00% 102.67%
722.palm_r 100.00% 100.00% 98.53% 100.00% 98.53%
731.astcenc_r 100.00% 100.63% 99.58% 100.00% 98.96%
736.ocio_r 100.00% 98.95% 100.00% 100.00% 101.06%
737.gmsh_r 100.00% 102.01% 102.01% 100.00% 100.00%
748.flightdm_r 100.00% 99.20% 99.20% 100.00% 100.00%
749.fotonik3d_r 100.00% 98.46% 99.12% 100.00% 100.67%
765.roms_r 100.00% 100.00% 98.70% 100.00% 98.70%
766.femflow_r 100.00% 97.79% 96.69% 100.00% 98.87%
767.nest_r 100.00% 118.99% 121.52% 100.00% 102.13%
772.marian_r 100.00% 101.70% 101.28% 100.00% 99.58%
782.lbm_r 100.00% 99.35% 98.92% 100.00% 99.56%
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
709.cactus_r 100.00% 102.61% 103.56% 100.00% 100.93%
722.palm_r 100.00% 102.86% 102.45% 100.00% 99.60%
731.astcenc_r 100.00% 99.03% 96.83% 100.00% 97.78%
736.ocio_r 100.00% 100.58% 100.66% 100.00% 100.08%
737.gmsh_r 100.00% 101.51% 100.96% 100.00% 99.46%
748.flightdm_r 100.00% 102.23% 101.15% 100.00% 98.94%
749.fotonik3d_r 100.00% 105.26% 104.59% 100.00% 99.36%
765.roms_r 100.00% 101.99% 100.85% 100.00% 98.89%
766.femflow_r 100.00% 102.23% 102.51% 100.00% 100.27%
767.nest_r 100.00% 100.65% 101.08% 100.00% 100.43%
772.marian_r 100.00% 100.61% 100.89% 100.00% 100.29%
782.lbm_r 100.00% 100.00% 100.51% 100.00% 100.51%
Smaller binary sizes are always better.