SPEC CPU result comparison

15th October 2026 - Normalized report

SPECINT 2006: LTO generic -O2 - Released GCCs vs Trunk

-O2 -flto

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
400.perlbench 100.00% 102.81% 101.73% 100.00% 98.95%
401.bzip2 100.00% 98.97% 98.97% 100.00% 100.00%
403.gcc 100.00% 100.00% 100.00% 100.00% 100.00%
429.mcf 100.00% 100.00% 100.67% 100.00% 100.67%
445.gobmk 100.00% 98.33% 98.33% 100.00% 100.00%
456.hmmer 100.00% 101.35% 111.69% 100.00% 110.20%
458.sjeng 100.00% 97.71% 98.12% 100.00% 100.43%
462.libquantum 100.00% 101.49% 91.96% 100.00% 90.62%
464.h264ref 100.00% 99.39% 98.79% 100.00% 99.39%
471.omnetpp 100.00% 99.15% 98.29% 100.00% 99.14%
473.astar 100.00% 98.29% 97.64% 100.00% 99.35%
483.xalancbmk 100.00% 100.00% 96.68% 100.00% 96.68%
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
400.perlbench 100.00% 102.89% 103.02% 100.00% 100.12%
401.bzip2 100.00% 105.74% 104.51% 100.00% 98.83%
403.gcc 100.00% 101.34% 101.34% 100.00% 100.00%
429.mcf 100.00% 101.06% 101.30% 100.00% 100.24%
445.gobmk 100.00% 100.30% 100.31% 100.00% 100.01%
456.hmmer 100.00% 102.27% 101.79% 100.00% 99.52%
458.sjeng 100.00% 100.09% 100.11% 100.00% 100.02%
462.libquantum 100.00% 98.83% 96.29% 100.00% 97.42%
464.h264ref 100.00% 100.92% 100.75% 100.00% 99.83%
471.omnetpp 100.00% 106.76% 106.69% 100.00% 99.94%
473.astar 100.00% 102.53% 101.98% 100.00% 99.46%
483.xalancbmk 100.00% 115.19% 115.00% 100.00% 99.84%
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
400.perlbench 100.00% 98.71% 101.94% 100.00% 103.27%
401.bzip2 100.00% 100.00% 99.16% 100.00% 99.16%
403.gcc 100.00% 100.00% 100.00% 100.00% 100.00%
429.mcf 100.00% 99.13% 99.13% 100.00% 100.00%
445.gobmk 100.00% 100.40% 98.99% 100.00% 98.59%
456.hmmer 100.00% 100.00% 101.03% 100.00% 101.03%
458.sjeng 100.00% 98.59% 99.60% 100.00% 101.02%
462.libquantum 100.00% 97.96% 98.87% 100.00% 100.93%
464.h264ref 100.00% 100.00% 98.79% 100.00% 98.79%
471.omnetpp 100.00% 99.36% 102.14% 100.00% 102.80%
473.astar 100.00% 99.55% 99.55% 100.00% 100.00%
483.xalancbmk 100.00% 95.08% 96.25% 100.00% 101.23%
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
400.perlbench 100.00% 102.89% 103.02% 100.00% 100.13%
401.bzip2 100.00% 105.85% 104.62% 100.00% 98.83%
403.gcc 100.00% 101.35% 101.34% 100.00% 100.00%
429.mcf 100.00% 101.05% 101.30% 100.00% 100.24%
445.gobmk 100.00% 100.30% 100.31% 100.00% 100.01%
456.hmmer 100.00% 102.24% 101.78% 100.00% 99.55%
458.sjeng 100.00% 100.09% 100.12% 100.00% 100.03%
462.libquantum 100.00% 98.56% 96.02% 100.00% 97.43%
464.h264ref 100.00% 100.91% 100.75% 100.00% 99.84%
471.omnetpp 100.00% 106.76% 106.69% 100.00% 99.94%
473.astar 100.00% 102.53% 102.16% 100.00% 99.64%
483.xalancbmk 100.00% 115.18% 115.00% 100.00% 99.84%
Smaller binary sizes are always better.