# NVIDIA GeForce RTX 4080, 76 SMs, sm_89, iters=20000 reps=3
mode       best TF/s   all runs
f16/f16        214.2 193.2/214.2/213.8
f16/f32        107.0 106.9/107.0/107.0
bf16/f32       106.9 106.9/106.9/106.9
s8/s32         427.1 427.0/427.1/426.6
dp4a           106.0 106.0/106.0/105.9
ffma            50.8 50.7/50.8/50.8
