Earlier this day I did a livestream on my YouTube channel to try using an Nvidia GeForce GTX 1080 on a Raspberry Pi Compute Module 4.
Or no longer it is most attention-grabbing been about a hours, nonetheless I’ve already gotten right solutions for better debugging than I change into in a reveal to attain on the lunge. And somebody identified it is probably going to be the case, as a consequence of 32-bit memory barriers on the BCM2711’s PCIe bus, that no GPU with more than 4 GB of onboard RAM might perchance perchance work. Though it is laborious to substantiate there’d be no instrument workaround—even 1 and 2 GB graphics cards (AMD and Nvidia) are crashing the kernel in identical ways.
The corpulent livestream is obtainable on replay and is embedded beneath:
And in that livestream, I in contrast the ‘chonky’ size of the 1080 card to the miniature ASRock Rack M2_VGA card I tested about a weeks within the past, tried installing Nvidia’s proprietary driver on 64-bit Pi OS (which locked up), tried recompiling the kernel with Nouveau (which locked up), after which documented your entire failures within the GitHub issue linked earlier in this put up.
I change into staring at for failure, and mentioned as essential at the starting of the lunge—5% likelihood of success, 10% likelihood of releasing the magic smoke—nonetheless it completely would be good to salvage any graphics card to work effectively on a Pi, for a diversity of makes employ of. However doubtlessly no longer AAA gaming, since gaming on Linux is a elaborate proposition even within the categorical of circumstances.
Some folks also instructed testing on an Nvidia Jetson board, or a Honeycomb, nonetheless the passe is a piece costly, and the latter appears to be unobtanium straight away (as with so many parts at the second). I mean heck, I would also enjoy to secure a manner to mortgage an Ampere server for a handy guide a rough time so I might perchance perchance veritably employ the categorical ARM hardware on offer… nonetheless alas these costs are astronomically beyond my funds straight away 🙂
I could sustain plugging away, and with a piece luck if no longer me, then somebody will salvage an AMD or Nvidia GPU engaged on the Pi. Coreforge is simply discontinuance with an old trend Radeon 6450.