SPEC CPU result comparison

Comparison of x86_64 levels - Summer 2022

SPECFP 2006: LTO -Ofast - Different x86_64 levels

-flto -Ofast

AMD EPYC 7543P (zen3 based Milan)

Run-times

Benchmark x86-64 x86-64-v2 x86-64-v3 x86-64-v2 x86-64-v3
410.bwaves 100.00% 100.00% 98.87% 100.00% 98.87%
416.gamess 100.00% 99.20% 97.79% 100.00% 98.58%
433.milc 100.00% 100.00% 105.75% 100.00% 105.75%
434.zeusmp 100.00% 100.00% 80.50% 100.00% 80.50%
435.gromacs 100.00% 100.00% 91.59% 100.00% 91.59%
436.cactusADM 100.00% 103.28% 71.53% 100.00% 69.26%
437.leslie3d 100.00% 100.76% 86.11% 100.00% 85.46%
444.namd 100.00% 98.99% 89.34% 100.00% 90.24%
447.dealII 100.00% 99.34% 100.00% 100.00% 100.66%
450.soplex 100.00% 99.19% 100.00% 100.00% 100.82%
453.povray 100.00% 100.00% 95.76% 100.00% 95.76%
454.calculix 100.00% 100.00% 82.32% 100.00% 82.32%
459.GemsFDTD 100.00% 100.00% 94.15% 100.00% 94.15%
465.tonto 100.00% 100.00% 96.34% 100.00% 96.34%
470.lbm 100.00% 100.61% 76.97% 100.00% 76.51%
481.wrf 100.00% 100.00% 82.05% 100.00% 82.05%
482.sphinx3 100.00% 98.66% 96.51% 100.00% 97.83%
The compiler used was GCC 12.1.0.
Naturally, the smaller the run-times, the better. On the other hand, rates and especially their geometric means are better when they are greater.
See also continuous gcc tuning benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/tuning.

Non-debug executable sizes

Click to display/hide
Benchmark x86-64 x86-64-v2 x86-64-v3 x86-64-v2 x86-64-v3
410.bwaves 100.00% 99.65% 90.06% 100.00% 90.38%
416.gamess 100.00% 99.99% 100.10% 100.00% 100.11%
433.milc 100.00% 103.93% 102.57% 100.00% 98.69%
434.zeusmp 100.00% 100.00% 100.01% 100.00% 100.01%
435.gromacs 100.00% 103.33% 103.91% 100.00% 100.56%
436.cactusADM 100.00% 99.46% 106.64% 100.00% 107.22%
437.leslie3d 100.00% 99.35% 113.13% 100.00% 113.87%
444.namd 100.00% 99.81% 98.51% 100.00% 98.70%
447.dealII 100.00% 99.57% 103.90% 100.00% 104.35%
450.soplex 100.00% 100.13% 104.35% 100.00% 104.21%
453.povray 100.00% 99.50% 96.90% 100.00% 97.38%
454.calculix 100.00% 99.14% 106.42% 100.00% 107.35%
459.GemsFDTD 100.00% 100.41% 115.98% 100.00% 115.51%
465.tonto 100.00% 99.44% 104.23% 100.00% 104.82%
470.lbm 100.00% 99.89% 91.72% 100.00% 91.83%
481.wrf 100.00% 99.88% 107.94% 100.00% 108.07%
482.sphinx3 100.00% 99.97% 104.78% 100.00% 104.81%
Smaller binary sizes are always better.

Intel Cascade Lake Xeon

Run-times

Benchmark x86-64 x86-64-v2 x86-64-v3 x86-64-v2 x86-64-v3
410.bwaves 100.00% 100.00% 99.43% 100.00% 99.43%
416.gamess 100.00% 99.80% 95.97% 100.00% 96.16%
433.milc 100.00% 98.57% 92.43% 100.00% 93.78%
434.zeusmp 100.00% 100.81% 80.76% 100.00% 80.11%
435.gromacs VE 100.00% 84.71%
436.cactusADM 100.00% 97.54% 64.75% 100.00% 66.39%
437.leslie3d 100.00% 100.52% 87.86% 100.00% 87.40%
444.namd 100.00% 100.00% 84.54% 100.00% 84.54%
447.dealII 100.00% 98.09% 96.82% 100.00% 98.70%
450.soplex 100.00% 99.39% 100.81% 100.00% 101.43%
453.povray 100.00% 100.00% 95.16% 100.00% 95.16%
454.calculix 100.00% 100.43% 68.97% 100.00% 68.67%
459.GemsFDTD 100.00% 100.65% 93.55% 100.00% 92.95%
465.tonto 100.00% 99.48% 97.16% 100.00% 97.67%
470.lbm 100.00% 100.00% 99.53% 100.00% 99.53%
481.wrf 100.00% 100.74% 81.55% 100.00% 80.95%
482.sphinx3 100.00% 96.83% 90.98% 100.00% 93.95%
The compiler used was GCC 12.1.0.
Naturally, the smaller the run-times, the better. On the other hand, rates and especially their geometric means are better when they are greater.
See also continuous gcc tuning benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/tuning.

Non-debug executable sizes

Click to display/hide
Benchmark x86-64 x86-64-v2 x86-64-v3 x86-64-v2 x86-64-v3
410.bwaves 100.00% 99.65% 90.06% 100.00% 90.38%
416.gamess 100.00% 99.99% 100.10% 100.00% 100.11%
433.milc 100.00% 103.93% 102.57% 100.00% 98.69%
434.zeusmp 100.00% 100.00% 100.01% 100.00% 100.01%
435.gromacs VE 100.00% 100.56%
436.cactusADM 100.00% 99.46% 106.64% 100.00% 107.22%
437.leslie3d 100.00% 99.35% 113.13% 100.00% 113.87%
444.namd 100.00% 99.81% 98.51% 100.00% 98.70%
447.dealII 100.00% 99.57% 103.90% 100.00% 104.35%
450.soplex 100.00% 100.13% 104.35% 100.00% 104.21%
453.povray 100.00% 99.50% 96.90% 100.00% 97.38%
454.calculix 100.00% 99.14% 106.42% 100.00% 107.35%
459.GemsFDTD 100.00% 100.41% 115.98% 100.00% 115.51%
465.tonto 100.00% 99.44% 104.23% 100.00% 104.82%
470.lbm 100.00% 99.89% 91.72% 100.00% 91.83%
481.wrf 100.00% 99.88% 107.94% 100.00% 108.07%
482.sphinx3 100.00% 99.97% 104.78% 100.00% 104.81%
Smaller binary sizes are always better.