SPEC CPU result comparison

15th October 2026 - Normalized report

SPEC2026 INTrate: Standard -Ofast native - LLVM, AOCC and GCC

-Ofast -march=native

Please note that even though the header of this comparison says -Ofast, benchmarks 710.omnetpp_r, 734.vpr_r, 735.gem5_r and 753.ns3_r were additionally compiled with -fno-finite-math-only.

AMD EPYC 8635P (zen5 based Sorano)

Run-times

Benchmark LLVM AOCC GCC AOCC GCC
706.stockfish_r 100.00% 100.84% 88.24% 100.00% 87.50%
707.ntest_r 100.00% 94.74% 97.03% 100.00% 102.42%
708.sqlite_r 100.00% 100.83% 95.03% 100.00% 94.25%
710.omnetpp_r 100.00% 102.76% 92.57% 100.00% 90.08%
714.cpython_r 100.00% 100.00% 92.74% 100.00% 92.74%
721.gcc_r 100.00% 99.39% 97.57% 100.00% 98.17%
723.llvm_r 100.00% 100.64% 100.00% 100.00% 99.37%
727.cppcheck_r 100.00% 102.18% 98.91% 100.00% 96.79%
729.abc_r 100.00% 102.71% 99.10% 100.00% 96.48%
734.vpr_r 100.00% 107.34% 95.25% 100.00% 88.73%
735.gem5_r 100.00% 99.11% 95.99% 100.00% 96.85%
750.sealcrypto_r 100.00% 228.11% 84.32% 100.00% 36.97%
753.ns3_r 100.00% 100.00% 98.96% 100.00% 98.96%
777.zstd_r 100.00% 109.85% 101.97% 100.00% 92.83%
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
706.stockfish_r 100.00% 99.55% 102.73% 100.00% 103.20%
707.ntest_r 100.00% 97.37% 99.42% 100.00% 102.11%
708.sqlite_r 100.00% 95.24% 99.49% 100.00% 104.46%
710.omnetpp_r 100.00% 98.38% 104.35% 100.00% 106.07%
714.cpython_r 100.00% 98.57% 102.88% 100.00% 104.38%
721.gcc_r 100.00% 98.58% 102.71% 100.00% 104.18%
723.llvm_r 100.00% 99.29% 128.09% 100.00% 129.00%
727.cppcheck_r 100.00% 98.48% 112.54% 100.00% 114.29%
729.abc_r 100.00% 94.00% 126.07% 100.00% 134.12%
734.vpr_r 100.00% 94.94% 108.03% 100.00% 113.79%
735.gem5_r 100.00% 99.38% 138.82% 100.00% 139.68%
750.sealcrypto_r 100.00% 94.27% 130.16% 100.00% 138.08%
753.ns3_r 100.00% 97.45% 112.20% 100.00% 115.14%
777.zstd_r 100.00% 87.75% 140.02% 100.00% 159.57%
Smaller binary sizes are always better.