SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2017 INTrate: 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
500.perlbench_r 100.00% 100.67% 100.67% 100.00% 100.00%
502.gcc_r 100.00% 99.17% 98.34% 100.00% 99.16%
505.mcf_r 100.00% 100.95% 100.00% 100.00% 99.06%
520.omnetpp_r 100.00% 98.47% 101.53% 100.00% 103.10%
523.xalancbmk_r 100.00% 100.89% 100.00% 100.00% 99.12%
525.x264_r 100.00% 100.00% 102.31% 100.00% 102.31%
531.deepsjeng_r 100.00% 98.15% 98.94% 100.00% 100.81%
541.leela_r 100.00% 97.58% 98.46% 100.00% 100.90%
548.exchange2_r 100.00% 97.77% 91.50% 100.00% 93.58%
557.xz_r 100.00% 101.07% 101.49% 100.00% 100.42%
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
500.perlbench_r 100.00% 102.32% 103.15% 100.00% 100.82%
502.gcc_r 100.00% 103.27% 103.18% 100.00% 99.92%
505.mcf_r 100.00% 103.09% 102.96% 100.00% 99.87%
520.omnetpp_r 100.00% 103.65% 103.93% 100.00% 100.27%
523.xalancbmk_r 100.00% 110.70% 110.59% 100.00% 99.91%
525.x264_r 100.00% 102.60% 102.15% 100.00% 99.56%
531.deepsjeng_r 100.00% 100.00% 99.99% 100.00% 99.99%
541.leela_r 100.00% 101.15% 101.39% 100.00% 100.23%
548.exchange2_r 100.00% 100.47% 100.16% 100.00% 99.69%
557.xz_r 100.00% 103.25% 102.31% 100.00% 99.09%
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
500.perlbench_r 100.00% 98.60% 98.60% 100.00% 100.00%
502.gcc_r 100.00% 100.00% 100.00% 100.00% 100.00%
505.mcf_r 100.00% 101.44% 100.48% 100.00% 99.06%
520.omnetpp_r 100.00% 97.53% 97.94% 100.00% 100.42%
523.xalancbmk_r 100.00% 97.41% 96.76% 100.00% 99.33%
525.x264_r 100.00% 101.65% 102.89% 100.00% 101.22%
531.deepsjeng_r 100.00% 97.90% 96.26% 100.00% 98.33%
541.leela_r 100.00% 97.71% 97.51% 100.00% 99.79%
548.exchange2_r 100.00% 91.73% 88.31% 100.00% 96.26%
557.xz_r 100.00% 99.58% 99.58% 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
500.perlbench_r 100.00% 102.32% 103.15% 100.00% 100.82%
502.gcc_r 100.00% 103.27% 103.18% 100.00% 99.92%
505.mcf_r 100.00% 103.07% 102.60% 100.00% 99.54%
520.omnetpp_r 100.00% 103.65% 103.93% 100.00% 100.27%
523.xalancbmk_r 100.00% 110.70% 110.59% 100.00% 99.91%
525.x264_r 100.00% 102.60% 102.14% 100.00% 99.55%
531.deepsjeng_r 100.00% 100.00% 99.99% 100.00% 99.99%
541.leela_r 100.00% 101.15% 101.39% 100.00% 100.23%
548.exchange2_r 100.00% 100.47% 100.16% 100.00% 99.69%
557.xz_r 100.00% 103.25% 102.31% 100.00% 99.09%
Smaller binary sizes are always better.