I have seen some many examples of connecting an external GPU to the PI but I want know if the reverse is possible whereby the RPI GPU can be accessed via a PC. The use case is to use the PI as a poor man's BMC for a homelab server. In particular I am interested in a low power ryzen server in a 1U rack where the only PCIE space is occupied by a network card.
-
Which OS runs on your Ryzen server? – Janka Oct 4 '18 at 21:46
-
Linux based hypervisor (e.g Proxmox) but happy to do a full bare metal install of Ubuntu server or similar. – Parnesh Raniga Oct 4 '18 at 23:39
-
2This is a basic function of the X11 windowing system of your Linux OS on the server. See headless X or terminal server X. You can access it through the network with almost any computer then, it's not specific to the Raspberry Pi. – Janka Oct 5 '18 at 0:56
The Raspberry Pi is a expensive choice for the basis of a IPMI/BMC device, especially considering the whole point of IPMI is that it works without and completely seperate from the server OS, without configuration and drivers. Specifically, Pi-based IPMI devices need extra hardware because of the missing consumer-grade video input: https://github.com/Fmstrat/diy-ipmi
In stark contrast to IPMI devices, which can interact with BIOS and UEFI effortlessly, the Pi can never become a IBM-compatible GPU since no BIOS/UEFI on the planet will recognize it as one.
Last but not least, not a single Raspberry Pi does support PCI/Thunderbolt or AGP - at all.