SPEC CPU result comparison

15th October 2026 - Normalized report

SPECINT 2006: PGO+LTO -Ofast native - LLVM, AOCC and GCC

-Ofast -march=native -mtune=native -flto and -fprofile-instr-use vs. -profile-use

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark LLVM GCC
400.perlbench 100.00% 81.19%
401.bzip2 100.00% 97.74%
403.gcc 100.00% 92.03%
429.mcf 100.00% 92.70%
445.gobmk FE
456.hmmer 100.00% 52.44%
458.sjeng 100.00% 101.03%
462.libquantum 100.00% 66.87%
464.h264ref 100.00% 86.32%
471.omnetpp 100.00% 91.11%
473.astar 100.00% FE
483.xalancbmk 100.00% 82.66%
LLVM clang version 23.1.3 (/labs/mjambor/git/llvm 0d261d1ca552c95a8f007e061c787ac7132fbcbc), flang version 23.1.3 (/labs/mjambor/git/llvm 0d261d1ca552c95a8f007e061c787ac7132fbcbc)
AOCC AMD clang version 17.0.6 (CLANG: AOCC_5.2.0-Build#2035 2026_04_10)
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% 98.42%
401.bzip2 100.00% 65.33%
403.gcc 100.00% 88.92%
429.mcf 100.00% 74.81%
445.gobmk FE
456.hmmer 100.00% 86.52%
458.sjeng 100.00% 99.36%
462.libquantum 100.00% 70.70%
464.h264ref 100.00% 70.77%
471.omnetpp 100.00% 84.82%
473.astar 100.00% FE
483.xalancbmk 100.00% 82.07%
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM GCC
400.perlbench 100.00% 81.74%
401.bzip2 100.00% 96.66%
403.gcc 100.00% 89.58%
429.mcf 100.00% 94.29%
445.gobmk FE
456.hmmer 100.00% 43.00%
458.sjeng 100.00% 106.30%
462.libquantum 100.00% 88.04%
464.h264ref 100.00% 95.72%
471.omnetpp 100.00% 90.95%
473.astar 100.00% FE
483.xalancbmk 100.00% 84.04%
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% 104.72%
401.bzip2 100.00% 68.40%
403.gcc 100.00% 89.38%
429.mcf 100.00% 72.16%
445.gobmk FE
456.hmmer 100.00% 88.31%
458.sjeng 100.00% 99.34%
462.libquantum 100.00% 75.17%
464.h264ref 100.00% 71.70%
471.omnetpp 100.00% 85.66%
473.astar 100.00% FE
483.xalancbmk 100.00% 83.82%
Smaller binary sizes are always better.