# NVIDIA GeForce RTX 4090, 128 SMs, sm_89, iters=20000 reps=5
mode       best TF/s   all runs
f16/f16        320.6 305.2/292.8/294.7/320.6/317.9
f16/f32        160.4 159.8/160.4/159.8/159.7/160.4
bf16/f32       159.7 159.2/159.7/156.4/159.6/156.6
s8/s32         641.1 641.1/640.5/640.2/640.2/640.2
dp4a           159.1 159.0/159.1/159.1/159.1/158.9
ffma            76.4 76.4/73.0/76.4/76.3/76.3
