SPEC CPU result comparison

15th October 2026 - Normalized report

SPECINT 2006: 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
400.perlbench 100.00% 83.03%
401.bzip2 100.00% 98.14%
403.gcc 100.00% 96.60%
429.mcf 100.00% 89.74%
445.gobmk FE
456.hmmer 100.00% 69.10%
458.sjeng 100.00% 97.76%
462.libquantum 100.00% 121.35%
464.h264ref 100.00% 86.95%
471.omnetpp 100.00% 106.02%
473.astar 100.00% 93.93%
483.xalancbmk 100.00% 87.79%
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% 104.66%
401.bzip2 100.00% 74.54%
403.gcc 100.00% 81.48%
429.mcf 100.00% 89.78%
445.gobmk FE
456.hmmer 100.00% 92.87%
458.sjeng 100.00% 99.41%
462.libquantum 100.00% 72.68%
464.h264ref 100.00% 85.59%
471.omnetpp 100.00% 81.28%
473.astar 100.00% 77.96%
483.xalancbmk 100.00% 81.52%
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM GCC
400.perlbench 100.00% 83.01%
401.bzip2 100.00% 96.03%
403.gcc 100.00% 94.68%
429.mcf 100.00% 92.86%
445.gobmk FE
456.hmmer 100.00% 63.72%
458.sjeng 100.00% 104.96%
462.libquantum 100.00% 106.48%
464.h264ref 100.00% 91.90%
471.omnetpp 100.00% 97.88%
473.astar 100.00% 87.40%
483.xalancbmk 100.00% 83.03%
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% 104.66%
401.bzip2 100.00% 74.64%
403.gcc 100.00% 81.48%
429.mcf 100.00% 90.09%
445.gobmk FE
456.hmmer 100.00% 92.91%
458.sjeng 100.00% 99.41%
462.libquantum 100.00% 72.66%
464.h264ref 100.00% 85.59%
471.omnetpp 100.00% 81.29%
473.astar 100.00% 78.00%
483.xalancbmk 100.00% 81.68%
Smaller binary sizes are always better.