Using the built-in Intel GPU?

When I run nvtop, I see two GPU devices: a built-in Intel GPU (Iris Xe), and a RTX 3060.

image

image

Nx is working with the RTX 3060, but the Intel device is unused. Does Nx support the Intel GPU? Would it be possible to split the processing load across the two GPUs??

This is pretty interesting, can you swap between them at will for your OS without reboots? Back in the day you would have to choose in the BIOS which one runs.

I’m not sure. My server OS is Ubuntu 22.04. I haven’t run nvtop for a year. I didn’t notice the Intel GPU in the past. I guess the Intel GPU display came with a nvtop upgrade.

Both devices appear on the pci bus…

> lspci  | grep -i vga
00:02.0 VGA compatible controller: Intel Corporation TigerLake-LP GT2 [Iris Xe Graphics] (rev 01)
04:00.0 VGA compatible controller: NVIDIA Corporation GA106 [GeForce RTX 3060 Lite Hash Rate] (rev a1)

Very interesting, it makes complete sense as running a server(on a linux server distribution) with multiple GPUs would make it mandatory to hot-swap them and use a specific one.

I think the question is whether Nx GPU backend supports specifying a device that should be used.