# NVIDIA GeForce RTX 4080 SUPER, 80 SMs, sm_89, iters=20000 reps=3
mode       best TF/s   all runs
f16/f16        210.4 199.2/204.1/210.4
f16/f32        209.7 209.7/209.4/209.4
bf16/f32       209.3 209.3/208.5/208.4
s8/s32         417.0 417.0/417.0/417.0
dp4a           103.8 103.8/103.8/103.8
ffma            49.7 49.7/49.6/49.5
