UART, Serial, SPI now all in use and I have like 5 GPIO pins left. Next model I need 80 GPIOs ;)
I want to also suggest that another change in silicon would be unusually useful, related to the GPIO: Either add several smaller user Cortex-R5s, for GPIO in groups of 8, or a couple of much higher performance Cortex-R52 for larger blocks of GPIO; a single Cortex-R82 could do astonishing things if it ran its own hard realtime by end users without losing the APE or ISP. Something like an R52 or R82, in combination with a few smaller R5s, and a dedicated communications pipe between that and the rest of the standard Jetson content, would mean precision control on levels only available in very expensive custom silicon (and although it would cost more than existing Jetsons, it would still be far below the cost that prohibits high end hobbyists, or those learning hard realtime, and would make an interesting end product with a whole new market).
To put it in perspective, the fastest any drone can currently run on the best Jetson, without adding other realtime control systems, is something that drives relatively slowly on the ground, or when flying low to the ground, not very fast. Having even Cortex-R5 dedicated to I/O (which is for users) would dramatically improve the speeds at which a Jetson could adjust flight paths without other hardware.
Combine it with an programmable IO-APIC, and this would be a two generation leap using existing technology (although it would require integrating it into hardware at the silicon level). I really hope this gets forwarded to the silicon designers (hint: Eric and Winnie).
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.