is there a way to set swarmUI to use a specificGPU?

[deathtome1998]

my use case may not the typical use case as far as I can tell for swarmUI.

I'm planning to run it alongside an AI RP frontend called Lumiverse, and it hooks into some of the image gen progams, the two I know of for sure are swarmUI, and I think comfyUI.

which comfyUI is the stuff of my nightmares to try to figure out, so I was recommended to use swarmUI.

but I have two GPU's, one of which is in an external enclosure, and it's connected to my PC, and the only thing I run on it currently, is either splitting a bigger 70B LLM when I do local models for RPing, or if I'm using a online model, like I've been mostly lately, I run an embedding model on it through a self hosted embedding server, and it uses extra VRAM to use that everytime it has to work after I send a message in the AI RP frontend.

so I want my swarmUI to only use my main GPU, which is a 4090, since it's not going to be fighting with a fluctuating amount of VRAM usage from both the image model I use, and the embedding model, getting used.

but I don't know if there's anywhere to set that up, I haven't come across somewhere.

but swarmUI keeps trying to run the image model on the external GPU where I don't want it to run.

[mcmonkey4eva]

Relevant doc https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Using%20More%20GPUs.md

Main thing for your case is just server->backends->gpu_id on the comfy backend

if you have two GPUs then the correct ID is either 0 or 1