# NVIDIA GeForce RTX 4070 SUPER, 56 SMs, sm_89, iters=20000 reps=3
mode       best TF/s   all runs
f16/f16        153.6 143.4/150.3/153.6
f16/f32        153.6 153.6/153.6/153.6
bf16/f32       153.6 153.6/153.6/153.6
s8/s32         307.3 307.3/307.3/307.3
dp4a            76.5 76.5/76.5/76.5
ffma            36.7 36.7/36.7/36.7
