SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 FPrate: PGO -Ofast native - LLVM and GCC

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

Please note that event though the header of this comparison says -Ofast, benchmark 767.nest_r was additionally compiled with -fno-finite-math-only. Furthermore, benchmarks 737.gmsh_r and 748.flightdm_r were compiled with -fno-fast-math, so a big chunk of -Ofast was in fact disabled. (This might put GCC at a disadvantage compared to LLVM-based compilers because the latter defaults to -fno-math-errno and -fno-trapping-math, in future measurements, we will try to enable these flags.)

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark LLVM GCC
709.cactus_r 100.00% 103.61%
722.palm_r CE
731.astcenc_r 100.00% 138.46%
736.ocio_r 100.00% 96.39%
737.gmsh_r 100.00% 92.81%
748.flightdm_r 100.00% 88.87%
749.fotonik3d_r CE
765.roms_r CE
766.femflow_r 100.00% CE
767.nest_r 100.00% 89.81%
772.marian_r 100.00% 62.50%
782.lbm_r 100.00% 41.61%
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
709.cactus_r 100.00% 95.70%
722.palm_r CE
731.astcenc_r 100.00% 111.79%
736.ocio_r 100.00% 81.68%
737.gmsh_r 100.00% 83.18%
748.flightdm_r 100.00% 99.26%
749.fotonik3d_r CE
765.roms_r CE
766.femflow_r 100.00% CE
767.nest_r 100.00% 86.02%
772.marian_r 100.00% 95.26%
782.lbm_r 100.00% 48.03%
Smaller binary sizes are always better.