Replies: 2 comments
-
That is indeed the best solution - right now the image gen always uses the first dGPU in a multigpu setup. |
Beta Was this translation helpful? Give feedback.
0 replies
-
Another option would be to spin up an external SD UI in API mode (Like Forge-ZLUDA) running on the other GPU. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
I have 2 discrete GPUs in my system:
Whenever I try loading both a LLM and a sd model with koboldcpp, both models get loaded on the same GPU. I would like to have LLM running on the 6800 and sd running on the 5700 XT, but I can't figure out if this is even possible.
Only solution I found was to have two instances of koboldcpp, one for text only and the other for image generation only.
Beta Was this translation helpful? Give feedback.
All reactions