SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 INTrate: PGO+LTO native -Ofast - Released GCCs vs Trunk

-Ofast -flto -march=native -mtune=native -fprofile-use

Please note that even though the header of this comparison says -Ofast, benchmarks 710.omnetpp_r, 734.vpr_r, 735.gem5_r and 753.ns3_r were additionally compiled with -fno-finite-math-only.

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark GCC 15 GCC 16 Trunk GCC 16 Trunk
706.stockfish_r 100.00% 72.19% 68.87% 100.00% 95.41%
707.ntest_r 100.00% 100.75% 100.75% 100.00% 100.00%
708.sqlite_r 100.00% 99.31% 98.40% 100.00% 99.08%
710.omnetpp_r 100.00% 95.68% 98.67% 100.00% 103.12%
714.cpython_r 100.00% 95.38% 98.55% 100.00% 103.33%
721.gcc_r 100.00% 100.00% 99.29% 100.00% 99.29%
723.llvm_r 100.00% 100.74% 98.26% 100.00% 97.54%
727.cppcheck_r 100.00% 98.75% 95.01% 100.00% 96.21%
729.abc_r 100.00% 98.31% 97.47% 100.00% 99.14%
734.vpr_r 100.00% 98.93% 98.93% 100.00% 100.00%
735.gem5_r 100.00% 100.00% 100.00% 100.00% 100.00%
750.sealcrypto_r 100.00% 80.89% 47.64% 100.00% 58.90%
753.ns3_r 100.00% 100.00% 100.00% 100.00% 100.00%
777.zstd_r 100.00% 91.70% 91.70% 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
706.stockfish_r 100.00% 102.93% 101.11% 100.00% 98.23%
707.ntest_r 100.00% 100.30% 100.08% 100.00% 99.78%
708.sqlite_r 100.00% 109.39% 104.46% 100.00% 95.49%
710.omnetpp_r 100.00% 102.44% 101.50% 100.00% 99.09%
714.cpython_r 100.00% 104.05% 103.76% 100.00% 99.72%
721.gcc_r 100.00% 109.89% 102.58% 100.00% 93.34%
723.llvm_r 100.00% 105.59% 101.62% 100.00% 96.24%
727.cppcheck_r 100.00% 103.82% 98.89% 100.00% 95.25%
729.abc_r 100.00% 101.09% 100.51% 100.00% 99.43%
734.vpr_r 100.00% 107.08% 104.30% 100.00% 97.40%
735.gem5_r 100.00% 102.43% 102.73% 100.00% 100.29%
750.sealcrypto_r 100.00% 96.89% 93.38% 100.00% 96.38%
753.ns3_r 100.00% 102.02% 101.07% 100.00% 99.06%
777.zstd_r 100.00% 103.19% 103.22% 100.00% 100.02%
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
706.stockfish_r 100.00% 77.78% 75.19% 100.00% 96.67%
707.ntest_r 100.00% 98.25% 98.25% 100.00% 100.00%
708.sqlite_r 100.00% 98.77% 95.81% 100.00% 97.01%
710.omnetpp_r 100.00% 97.19% 97.19% 100.00% 100.00%
714.cpython_r 100.00% 94.99% 97.05% 100.00% 102.17%
721.gcc_r 100.00% 99.27% 98.06% 100.00% 98.78%
723.llvm_r 100.00% 100.82% 98.64% 100.00% 97.84%
727.cppcheck_r 100.00% 99.27% 100.97% 100.00% 101.71%
729.abc_r 100.00% 96.29% 94.43% 100.00% 98.07%
734.vpr_r 100.00% 100.00% 100.00% 100.00% 100.00%
735.gem5_r 100.00% 99.32% 99.32% 100.00% 100.00%
750.sealcrypto_r 100.00% 83.03% 40.83% 100.00% 49.17%
753.ns3_r 100.00% 100.00% 98.53% 100.00% 98.53%
777.zstd_r 100.00% 91.86% 90.58% 100.00% 98.60%
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
706.stockfish_r 100.00% 102.83% 100.93% 100.00% 98.15%
707.ntest_r 100.00% 100.36% 100.14% 100.00% 99.78%
708.sqlite_r 100.00% 109.53% 104.41% 100.00% 95.33%
710.omnetpp_r 100.00% 102.34% 101.59% 100.00% 99.27%
714.cpython_r 100.00% 104.02% 103.71% 100.00% 99.70%
721.gcc_r 100.00% 110.01% 102.62% 100.00% 93.28%
723.llvm_r 100.00% 105.74% 101.72% 100.00% 96.20%
727.cppcheck_r 100.00% 103.89% 99.01% 100.00% 95.31%
729.abc_r 100.00% 101.08% 100.55% 100.00% 99.47%
734.vpr_r 100.00% 107.23% 104.41% 100.00% 97.37%
735.gem5_r 100.00% 102.59% 102.91% 100.00% 100.31%
750.sealcrypto_r 100.00% 97.36% 93.30% 100.00% 95.82%
753.ns3_r 100.00% 102.07% 101.16% 100.00% 99.10%
777.zstd_r 100.00% 102.44% 102.40% 100.00% 99.96%
Smaller binary sizes are always better.