Issue with corrupted image at the final stage when forcing CPU usage in Automatic1111 #16513
Replies: 2 comments 6 replies
-
I never seen this before
|
Beta Was this translation helpful? Give feedback.
-
sysinfo-2024-09-24-16-04.json its a raspberry 5 with a SWAP of 80GB on a NVME |
Beta Was this translation helpful? Give feedback.
-
Hi everyone,
I'm having an issue with Automatic1111 when forcing it to use the CPU with the --device cpu option. Everything seems to work fine at the beginning, but at the final stage of generation, the image becomes corrupted. I also enabled the --no-half option to avoid using float16 and stick to float32, but that didn’t solve the issue.
Has anyone encountered this problem before? Could it be related to VAE settings or something else I need to adjust when running on CPU only?
Thanks for your help!
Beta Was this translation helpful? Give feedback.
All reactions