Replies: 3 comments
-
now it works again, |
Beta Was this translation helpful? Give feedback.
0 replies
-
Its a memory leek that happens over time. I'm not sure what triggers it, but restarting Forge fixes it. Memory seems to become fractured and then each image after that is a bit slower and the spikes get more spikey... New.project.mp4 |
Beta Was this translation helpful? Give feedback.
0 replies
-
I've noticed weird things happens if you interrupt the generation of the image. Can you test if this is what triggers your issue? |
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
-
I updated to the latest and noticed that my sdXL models no longer ping my GPU at 100%, in stead its a series of spikes (3% then then 40% then 88% then 3% etc.) it doubles the image generation time. Switching to Flux1 and I get 100% utilization of the GPU.
I still have "webui_forge_cu121_torch231.zip" so I extracted and ran that build to test. XL runs at 100% on that version.
4070 12gb and 128gb of ram
Beta Was this translation helpful? Give feedback.
All reactions