Recently acquired 240 M3. After doing some general house keeping on it, I have been trying to passthrough 2 different GPUs in ESXi 6.7 (The Cisco release). The two cards are both Nvidia cards. One is a low end consumer card, and the other is a GTX 1050 Ti. While I know both of these are definitely unsupported by Nvidia (If passed through without doing some spoofing, the infamous Code 43 "bug" will occur.). While I know these should run into that issue,I was not expecting them to cause a full system panic/crash. I can have them set to pass through at any pint without issues, but as soon as I click "Start" on the VM with the card attached, I loose the web ui, and a short time later I can hear the fans maxing out, the PSOD reads "recursive panic on same cpu". There is a corresponding syslog entry, but I am away from my syslog server to put exactly what it had.
Is there something OPROM related that I need to configure? I will be moving to AMD GPUs (Fire Pro 1 Slots) s soon as I can, but I am afraid I will have the same crashes. CIMC does report that both GPUs are GPUs, and that makes me believe that this is a configuration issue. I do know that while my modified R710 still faced the Code 43, it never once crashed from a pass-through.
I love the Cisco ecosystem, and would love to fix this issue and keep using the USC platform over Dell and HPE.
Thanks!
