SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 FPrate: LTO -O3 x86-64-v3 - LLVM and GCC

-O3 -flto -march=x86-64-v3

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
709.cactus_r 100.00% 112.75%
722.palm_r 100.00% 147.30%
731.astcenc_r 100.00% 155.00%
736.ocio_r 100.00% 99.19%
737.gmsh_r 100.00% 92.69%
748.flightdm_r 100.00% 90.07%
749.fotonik3d_r 100.00% 105.93%
765.roms_r 100.00% 97.84%
766.femflow_r 100.00% 113.21%
767.nest_r 100.00% 104.81%
772.marian_r 100.00% 76.00%
782.lbm_r 100.00% 64.89%
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% 104.30%
722.palm_r 100.00% 89.74%
731.astcenc_r 100.00% 131.40%
736.ocio_r 100.00% 92.55%
737.gmsh_r 100.00% 99.93%
748.flightdm_r 100.00% 91.27%
749.fotonik3d_r 100.00% 38.59%
765.roms_r 100.00% 78.05%
766.femflow_r 100.00% 65.25%
767.nest_r 100.00% 88.74%
772.marian_r 100.00% 96.72%
782.lbm_r 100.00% 55.84%
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM GCC
709.cactus_r 100.00% 110.74%
722.palm_r 100.00% 144.23%
731.astcenc_r 100.00% 144.17%
736.ocio_r 100.00% 95.64%
737.gmsh_r 100.00% 93.91%
748.flightdm_r 100.00% 96.26%
749.fotonik3d_r 100.00% 100.52%
765.roms_r 100.00% 95.25%
766.femflow_r 100.00% 114.49%
767.nest_r 100.00% 102.26%
772.marian_r 100.00% 113.79%
782.lbm_r 100.00% 88.76%
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
709.cactus_r 100.00% 104.30%
722.palm_r 100.00% 89.74%
731.astcenc_r 100.00% 131.40%
736.ocio_r 100.00% 91.90%
737.gmsh_r 100.00% 99.96%
748.flightdm_r 100.00% 90.90%
749.fotonik3d_r 100.00% 38.60%
765.roms_r 100.00% 78.06%
766.femflow_r 100.00% 65.10%
767.nest_r 100.00% 89.23%
772.marian_r 100.00% 96.20%
782.lbm_r 100.00% 56.20%
Smaller binary sizes are always better.