After reading this article today in electrek:
First picture of Tesla’s new NVIDIA onboard supercomputer for Autopilot installed in a car
It seems that the current hw2 vehicles can realistically achieve level 4 with the processing power of the current chipset, and although a swap out for newer components is possible, Tesla believes it can fine tune the software to alleviate the need for a more powerful hardware swap. With this in mind, I wonder if this same techniques could be applicable to the previous generation hardware to achieve a higher level of autonomy with a limited hardware suite.
I'm always hoping for leaps in hw1 features but I'm thinking even if they could advance it, they would hold back until hw2 catches up to avoid the perception of broken promises and missed timelines. I can see a few folks getting angry after a hw2 update that brings it to parity, only to have hw1 gain another feature that hw2 has to wait for, even knowing that the day will come when hw2 passes hw1's capabilities.
First picture of Tesla’s new NVIDIA onboard supercomputer for Autopilot installed in a car
It seems that the current hw2 vehicles can realistically achieve level 4 with the processing power of the current chipset, and although a swap out for newer components is possible, Tesla believes it can fine tune the software to alleviate the need for a more powerful hardware swap. With this in mind, I wonder if this same techniques could be applicable to the previous generation hardware to achieve a higher level of autonomy with a limited hardware suite.
I'm always hoping for leaps in hw1 features but I'm thinking even if they could advance it, they would hold back until hw2 catches up to avoid the perception of broken promises and missed timelines. I can see a few folks getting angry after a hw2 update that brings it to parity, only to have hw1 gain another feature that hw2 has to wait for, even knowing that the day will come when hw2 passes hw1's capabilities.