stable-diffusion
stable-diffusion copied to clipboard
Crashes when running on 2nd GPU while a job is already running on 1st
If a txt2img job is running and using the 1st GPU, then I start a second job to use the 2nd GPU, that job crashes:
UNet: Running in eps-prediction mode
Killed
I'm running on a Linux based machine with several NVidia GTX 3080 GPUs. The single jobs run fine, but I'd like the ability to have several jobs running concurrently, each using their own GPU.
Any thoughts?
I have the same issue because of RAM, not because of GPU.