CUDA out of memory on automatic1111 Stable Diffusion

Asked Jul 03 '23 at 03:38

Active Aug 13 '23 at 01:49

Viewed 1,393 times

I'm running into an issue while running Stable Diffusion web UI AUTOMATIC1111. I have an RTX 2060 6GB, and I'm running into this issue.

OutOfMemoryError: CUDA out of memory. 
Tried to allocate 1.50 GiB 
 (GPU 0; 5.77 GiB total capacity; 3.34 GiB already allocated; 1.07 GiB free; 3.41 GiB reserved in total by PyTorch) 
If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. 
See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

Time taken: 57.59s

Torch active/reserved: 4200/4380 MiB, Sys VRAM: 5687/5910 MiB (96.23%)

I know it's a low amount of vram, but I didn't get this while running under Windows. I tried setting max_split_size_mb to 512mb

edited Aug 13 '23 at 01:49

karel

5,489
46
45
50

asked Jul 03 '23 at 03:38

Leirbag Ed Satierf

1

It's telling you what the problem is. You're trying to load a model that's too large for your GPU – Abirbhav G. Jul 03 '23 at 05:39
yeah i get that, how come it's fine on windows then? – Leirbag Ed Satierf Jul 03 '23 at 20:44
i just want to know if there's any optimizations that are no implemented yet on linux, so my switch of operating systems isn't as hellish as it is. – Leirbag Ed Satierf Jul 03 '23 at 20:45

CUDA out of memory on automatic1111 Stable Diffusion

0 Answers0