SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 INTrate: PGO+LTO -O2 generic - LLVM and GCC

-O2 -flto and -fprofile-instr-use vs. -profile-use

Please note that the same "optimization levels" mean different things to different compilers. For example, LLVM and LLVM-based compiler tend to be much more aggressive at -O2, often with the corresponding consequences to size of the generated binaries and compilation time. GCC is sometimes more aggressive at -O3. LLVM is also often by default less strict when it comes to standard compliance by default and performs optimizations GCC would only do with a special compilation option or at -Ofast. Keep this is mind when comparing different compilers.

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark LLVM GCC
706.stockfish_r 100.00% 42.82%
707.ntest_r 100.00% 108.24%
708.sqlite_r 100.00% 95.52%
710.omnetpp_r 100.00% 86.05%
714.cpython_r 100.00% 94.35%
721.gcc_r 100.00% 101.62%
723.llvm_r 100.00% 99.01%
727.cppcheck_r 100.00% 97.63%
729.abc_r 100.00% 104.33%
734.vpr_r 100.00% 95.57%
735.gem5_r 100.00% 101.92%
750.sealcrypto_r 100.00% 116.97%
753.ns3_r 100.00% 105.94%
777.zstd_r 100.00% 91.30%
LLVM clang version 23.1.3 (/labs/mjambor/git/llvm 0d261d1ca552c95a8f007e061c787ac7132fbcbc), flang version 23.1.3 (/labs/mjambor/git/llvm 0d261d1ca552c95a8f007e061c787ac7132fbcbc)
GCC GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.

Non-debug executable sizes

Benchmark LLVM GCC
706.stockfish_r 100.00% 95.80%
707.ntest_r 100.00% 90.68%
708.sqlite_r 100.00% 91.96%
710.omnetpp_r 100.00% 75.63%
714.cpython_r 100.00% 80.44%
721.gcc_r 100.00% 82.92%
723.llvm_r 100.00% 76.46%
727.cppcheck_r 100.00% 78.54%
729.abc_r 100.00% 80.13%
734.vpr_r 100.00% 80.68%
735.gem5_r 100.00% 76.33%
750.sealcrypto_r 100.00% 81.04%
753.ns3_r 100.00% 70.60%
777.zstd_r 100.00% 83.90%
Smaller binary sizes are always better.

AMD EPYC 75F3 (zen3 based high-frequency Milan)

Run-times

Benchmark LLVM GCC
706.stockfish_r 100.00% 40.44%
707.ntest_r 100.00% 107.82%
708.sqlite_r 100.00% 96.07%
710.omnetpp_r 100.00% 83.20%
714.cpython_r 100.00% 103.83%
721.gcc_r 100.00% 101.68%
723.llvm_r 100.00% 101.33%
727.cppcheck_r 100.00% 93.39%
729.abc_r 100.00% 101.91%
734.vpr_r 100.00% 95.08%
735.gem5_r 100.00% 107.38%
750.sealcrypto_r 100.00% 108.95%
753.ns3_r 100.00% 100.00%
777.zstd_r 100.00% 91.75%
LLVM clang version 23.1.3 (https://github.com/llvm/llvm-project.git 0d261d1ca552c95a8f007e061c787ac7132fbcbc), flang version 23.1.3 (https://github.com/llvm/llvm-project.git 0d261d1ca552c95a8f007e061c787ac7132fbcbc)
GCC GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.

Non-debug executable sizes

Benchmark LLVM GCC
706.stockfish_r 100.00% 95.81%
707.ntest_r 100.00% 90.69%
708.sqlite_r 100.00% 91.96%
710.omnetpp_r 100.00% 75.61%
714.cpython_r 100.00% 80.43%
721.gcc_r 100.00% 82.92%
723.llvm_r 100.00% 76.45%
727.cppcheck_r 100.00% 78.70%
729.abc_r 100.00% 80.13%
734.vpr_r 100.00% 80.86%
735.gem5_r 100.00% 76.38%
750.sealcrypto_r 100.00% 80.66%
753.ns3_r 100.00% 70.61%
777.zstd_r 100.00% 83.93%
Smaller binary sizes are always better.