SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 FPrate: Standard -Ofast native - LLVM, ICX and GCC

-Ofast -march=native

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 ICX GCC ICX GCC
709.cactus_r 100.00% 36.95% 106.43% 100.00% 288.04%
722.palm_r 100.00% 101.31% 114.05% 100.00% 112.58%
731.astcenc_r 100.00% 88.55% 116.03% 100.00% 131.03%
736.ocio_r 100.00% 100.00% 93.03% 100.00% 93.03%
737.gmsh_r 100.00% 88.61% 93.58% 100.00% 105.61%
748.flightdm_r 100.00% 87.78% 91.04% 100.00% 103.71%
749.fotonik3d_r 100.00% 107.51% 101.34% 100.00% 94.26%
765.roms_r 100.00% 202.08% 92.92% 100.00% 45.98%
766.femflow_r 100.00% 94.64% 82.14% 100.00% 86.79%
767.nest_r 100.00% 88.97% 90.00% 100.00% 101.15%
772.marian_r 100.00% 80.00% 60.00% 100.00% 75.00%
782.lbm_r 100.00% 66.76% 61.18% 100.00% 91.63%
LLVM clang version 23.1.3 (/labs/mjambor/git/llvm 0d261d1ca552c95a8f007e061c787ac7132fbcbc), flang version 23.1.3 (/labs/mjambor/git/llvm 0d261d1ca552c95a8f007e061c787ac7132fbcbc)
ICX Intel(R) oneAPI DPC++/C++ Compiler 2026.1.1 (2026.1.1.20260724), ifx (IFX) 2026.1.1 20260724
GCC GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.

Non-debug executable sizes

Benchmark LLVM ICX GCC ICX GCC
709.cactus_r 100.00% 146.54% 106.65% 100.00% 72.78%
722.palm_r 100.00% 80.68% 77.29% 100.00% 95.80%
731.astcenc_r 100.00% 116.18% 137.87% 100.00% 118.67%
736.ocio_r 100.00% 99.35% 121.76% 100.00% 122.56%
737.gmsh_r 100.00% 99.24% 107.35% 100.00% 108.17%
748.flightdm_r 100.00% 117.73% 119.86% 100.00% 101.81%
749.fotonik3d_r 100.00% 98.84% 38.12% 100.00% 38.57%
765.roms_r 100.00% 89.17% 68.46% 100.00% 76.78%
766.femflow_r 100.00% 111.82% 87.99% 100.00% 78.69%
767.nest_r 100.00% 99.04% 110.21% 100.00% 111.27%
772.marian_r 100.00% 98.18% 108.35% 100.00% 110.36%
782.lbm_r 100.00% 55.75% 52.40% 100.00% 93.99%
Smaller binary sizes are always better.

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

Run-times

Benchmark LLVM ICX GCC ICX GCC
709.cactus_r 100.00% 37.46% 99.15% 100.00% 264.66%
722.palm_r 100.00% 97.33% 114.24% 100.00% 117.38%
731.astcenc_r 100.00% 106.61% 102.40% 100.00% 96.06%
736.ocio_r 100.00% 95.56% 94.78% 100.00% 99.18%
737.gmsh_r 100.00% 90.66% 95.44% 100.00% 105.26%
748.flightdm_r 100.00% 92.80% 94.24% 100.00% 101.55%
749.fotonik3d_r 100.00% 102.65% 102.65% 100.00% 100.00%
765.roms_r 100.00% 118.69% 90.82% 100.00% 76.52%
766.femflow_r 100.00% 102.74% 100.00% 100.00% 97.33%
767.nest_r 100.00% 96.15% 99.23% 100.00% 103.20%
772.marian_r 100.00% 96.30% 107.41% 100.00% 111.54%
782.lbm_r 100.00% 131.19% 110.93% 100.00% 84.56%
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)
ICX Intel(R) oneAPI DPC++/C++ Compiler 2026.1.1 (2026.1.1.20260724), ifx (IFX) 2026.1.1 20260724
GCC GCC 17.0.0 20261015 (experimental) (revision 84348add236100e5f69a3e9ec4f7b1038bf2b83d)
Naturally, the smaller the run-times, the better.

Non-debug executable sizes

Benchmark LLVM ICX GCC ICX GCC
709.cactus_r 100.00% 144.14% 108.60% 100.00% 75.34%
722.palm_r 100.00% 88.97% 80.62% 100.00% 90.61%
731.astcenc_r 100.00% 121.60% 157.71% 100.00% 129.69%
736.ocio_r 100.00% 98.03% 127.04% 100.00% 129.59%
737.gmsh_r 100.00% 99.62% 109.03% 100.00% 109.45%
748.flightdm_r 100.00% 117.52% 119.77% 100.00% 101.92%
749.fotonik3d_r 100.00% 101.61% 38.31% 100.00% 37.70%
765.roms_r 100.00% 98.22% 70.83% 100.00% 72.11%
766.femflow_r 100.00% 123.08% 95.04% 100.00% 77.22%
767.nest_r 100.00% 98.03% 109.39% 100.00% 111.59%
772.marian_r 100.00% 96.15% 107.02% 100.00% 111.31%
782.lbm_r 100.00% 63.24% 68.65% 100.00% 108.55%
Smaller binary sizes are always better.