10 times the energy used and only one tenth the processing speed.
It is like compering a modern LCD high definition TV with a 26" B&W CRT circa 1970....
To be honest I thought it was pretty clear what I was saying, but let me spell it out more simply.
Tesla FSD has the potential, with robotaxis, to change the landscape of Vehicle ownership forever.
Today the hardware driving FSD is just about enough with smart programming, to replace a human, by emulating what a human does. We don't have lidar sensors, we have two eyes and a brain which has been trained
By increasing HW3 to triple "just enough", Tesla gains the ability to make the FSD comprehensive.
To do that you need the grunt to process images and make decisions. That is what we do and, for the most, we do it very well.
The key competitor to Tesla HW is Nvidia. Witnessed by the fact that Tesla used to use their hardware until Nvidia fell behind the needs of Tesla.
It has already been covered here several times, that the Nvidia offering uses too much power for too little ability. I didn't think I had to go there again.
Nvidia's response to HW3 is to announce a new chip which produces less ability than HW4, for close to 2 times the power consumption of HW3 and a road map to 5 times the ability of HW4 for 10 times the power consumption.
Put another way, using the projected Nvidia hardware, A tesla would need special cooling for the computer and would drive about 2.5 miles less, every hour, than with the Tesla computer.
Never mind the fact that the gap between Nvidia announcements and actual product is about 2 years, which puts Tesla ahead again.
All of this means that Tesla vertical integration puts them ahead of the market, again, where their competitors can't even buy a competitive computer, let alone have the ability to use it to beat Tesla.
Personally I thought that having the right technology to gain an unassailable advantage in your chosen market segment was relevant to Glory.
Apparently I'm wrong