SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 FPrate: PGO+LTO -Ofast native - LLVM, AOCC and GCC

-Ofast -march=native -mtune=native -flto 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 AOCC GCC AOCC GCC
709.cactus_r 100.00% 47.01% 110.36% 100.00% 234.75%
722.palm_r CE 100.00% 70.12%
731.astcenc_r 100.00% 106.92% 134.23% 100.00% 125.54%
736.ocio_r 100.00% 104.44% 95.56% 100.00% 91.49%
737.gmsh_r 100.00% 100.00% 94.07% 100.00% 94.07%
748.flightdm_r 100.00% 98.14% 83.53% 100.00% 85.11%
749.fotonik3d_r CE 100.00% 114.21%
765.roms_r CE 100.00% 92.24%
766.femflow_r 100.00% 112.24% 95.92% 100.00% 85.45%
767.nest_r 100.00% 110.83% 91.18% 100.00% 82.27%
772.marian_r 100.00% 112.50% 68.75% 100.00% 61.11%
782.lbm_r 100.00% 41.60% 41.19% 100.00% 99.01%
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 AOCC GCC AOCC GCC
709.cactus_r 100.00% 134.25% 90.49% 100.00% 67.41%
722.palm_r CE 100.00% 72.97%
731.astcenc_r 100.00% 101.81% 94.18% 100.00% 92.51%
736.ocio_r 100.00% 99.99% 82.12% 100.00% 82.12%
737.gmsh_r 100.00% 105.48% 74.71% 100.00% 70.83%
748.flightdm_r 100.00% 103.24% 100.11% 100.00% 96.97%
749.fotonik3d_r CE 100.00% 65.02%
765.roms_r CE 100.00% 87.37%
766.femflow_r 100.00% 98.84% 64.63% 100.00% 65.39%
767.nest_r 100.00% 98.77% 73.39% 100.00% 74.30%
772.marian_r 100.00% 101.86% 83.89% 100.00% 82.36%
782.lbm_r 100.00% 62.79% 40.59% 100.00% 64.66%
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM GCC
709.cactus_r 100.00% 98.04%
722.palm_r CE
731.astcenc_r 100.00% 114.81%
736.ocio_r 100.00% 90.70%
737.gmsh_r 100.00% 94.50%
748.flightdm_r 100.00% 87.05%
749.fotonik3d_r CE
765.roms_r CE
766.femflow_r 100.00% 118.75%
767.nest_r 100.00% 100.51%
772.marian_r 100.00% 106.90%
782.lbm_r 100.00% 109.18%
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% 92.94%
722.palm_r CE
731.astcenc_r 100.00% 96.97%
736.ocio_r 100.00% 82.77%
737.gmsh_r 100.00% 75.68%
748.flightdm_r 100.00% 100.14%
749.fotonik3d_r CE
765.roms_r CE
766.femflow_r 100.00% 66.76%
767.nest_r 100.00% 74.61%
772.marian_r 100.00% 84.82%
782.lbm_r 100.00% 69.06%
Smaller binary sizes are always better.