# NVIDIA GeForce RTX 4090, 128 SMs, sm_89, iters=20000 reps=5
mode       best TF/s   all runs
f16/f16        330.8 302.7/302.7/330.8/330.8/330.8
f16/f32        328.9 328.9/328.9/328.9/328.9/328.9
bf16/f32       328.7 327.9/327.1/328.7/328.7/328.7
s8/s32         657.7 654.8/657.6/657.7/657.6/657.6
dp4a           163.5 163.5/163.4/163.5/163.4/163.4
ffma            78.6 78.5/78.5/78.6/78.4/78.3
