SPEC CPU result comparison

15th October 2026 - Normalized report

SPECINT 2006: LTO -O2 generic - LLVM and GCC

-O2 -flto

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
400.perlbench 100.00% 94.77%
401.bzip2 100.00% 97.17%
403.gcc 100.00% 100.00%
429.mcf 100.00% 91.08%
445.gobmk RE
456.hmmer 100.00% 121.81%
458.sjeng 100.00% 100.43%
462.libquantum 100.00% 104.75%
464.h264ref 100.00% 106.07%
471.omnetpp 100.00% 97.46%
473.astar 100.00% 93.44%
483.xalancbmk 100.00% 79.92%
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
400.perlbench 100.00% 97.38%
401.bzip2 100.00% 72.82%
403.gcc 100.00% 91.41%
429.mcf 100.00% 96.80%
445.gobmk RE
456.hmmer 100.00% 91.76%
458.sjeng 100.00% 100.45%
462.libquantum 100.00% 87.62%
464.h264ref 100.00% 81.12%
471.omnetpp 100.00% 89.93%
473.astar 100.00% 86.80%
483.xalancbmk 100.00% 108.04%
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM GCC
400.perlbench 100.00% 98.34%
401.bzip2 100.00% 98.54%
403.gcc 100.00% 100.00%
429.mcf 100.00% 91.97%
445.gobmk RE
456.hmmer 100.00% 114.42%
458.sjeng 100.00% 105.31%
462.libquantum 100.00% 107.65%
464.h264ref 100.00% 104.69%
471.omnetpp 100.00% 96.57%
473.astar 100.00% 93.22%
483.xalancbmk 100.00% 87.82%
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
400.perlbench 100.00% 97.39%
401.bzip2 100.00% 72.92%
403.gcc 100.00% 91.41%
429.mcf 100.00% 96.86%
445.gobmk RE
456.hmmer 100.00% 91.79%
458.sjeng 100.00% 100.45%
462.libquantum 100.00% 87.57%
464.h264ref 100.00% 81.13%
471.omnetpp 100.00% 89.94%
473.astar 100.00% 86.99%
483.xalancbmk 100.00% 108.20%
Smaller binary sizes are always better.