SPEC CPU result comparison

15th October 2026 - Normalized report

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

-Ofast -march=native -flto

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% 45.38% 115.26% 100.00% 253.98%
722.palm_r 100.00% 166.90% 120.34% 100.00% 72.11%
731.astcenc_r 100.00% 108.78% 117.18% 100.00% 107.72%
736.ocio_r 100.00% 98.37% 94.04% 100.00% 95.59%
737.gmsh_r 100.00% 97.48% 93.28% 100.00% 95.69%
748.flightdm_r 100.00% 100.90% 92.78% 100.00% 91.95%
749.fotonik3d_r 100.00% 100.00% 102.68% 100.00% 102.68%
765.roms_r 100.00% 103.38% 93.25% 100.00% 90.20%
766.femflow_r 100.00% 112.24% 95.92% 100.00% 85.45%
767.nest_r 100.00% 108.77% 98.08% 100.00% 90.18%
772.marian_r 100.00% 128.57% 64.29% 100.00% 50.00%
782.lbm_r 100.00% 67.84% 60.23% 100.00% 88.79%
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% 166.46% 95.13% 100.00% 57.15%
722.palm_r 100.00% 89.93% 82.60% 100.00% 91.85%
731.astcenc_r 100.00% 89.97% 138.64% 100.00% 154.10%
736.ocio_r 100.00% 99.74% 90.22% 100.00% 90.45%
737.gmsh_r 100.00% 98.97% 96.65% 100.00% 97.65%
748.flightdm_r 100.00% 104.79% 91.45% 100.00% 87.27%
749.fotonik3d_r 100.00% 33.47% 35.38% 100.00% 105.68%
765.roms_r 100.00% 65.15% 70.58% 100.00% 108.34%
766.femflow_r 100.00% 123.12% 67.62% 100.00% 54.93%
767.nest_r 100.00% 93.68% 84.99% 100.00% 90.72%
772.marian_r 100.00% 98.73% 93.60% 100.00% 94.80%
782.lbm_r 100.00% 114.37% 45.12% 100.00% 39.45%
Smaller binary sizes are always better.