Only thing I can imagine would be allowances for wider and deeper neural nets.
https://www.youtube.com/watch?v=-KxjVlaLBmk
The higher the speed of the system, the less it has to predict in the future and better can adapt to unforeseen situations. It's basically using the world as a model, for free.
Yeah I think this is it. My best guess, it's 200fps vs 2000fps for the same workload. As Tesla's data model gets larger and its computations get more complex, that 200fps will start to sag.
2000fps will provide a lot more wiggle room to do higher level computation on each frame.
36 / 200 frames = 18cm travel distance between each calculation.
So you either can calculate much more each 18cm or calculate shorter distances.
I think a shorter distance is not very useful but calculating more might be.
I think n = 8
...or more cameras, with better frame rates over a lot of cameras. My guess is that the fps rate is over all the cameras, so 10 cameras means going from 20fps per camera to 200fps per camera.
As a simple example in a different domain, years back I found that the best way to improve Tesseract's OCR accuracy was to ensure that I didn't feed it images at more than 150dpi because it would sometimes misrecognize dust, paper texture, etc. as characters.