You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is more pertinent to the worker use case but certain recent changes by ComfyUI and myself have highlighted that there could be some better finessing of memory conditions. While I do feel that it should be largely in the realm of ComfyUI internals to manage this, that project has changed or added a large amount of functionality which is harmful to the worker use case.
In particular the change in commit 4e99a01 highlights that there should be at least a 'high vram' mode which opts-in to either default comfy behavior or the deliberately more aggressive use of memory we occasionally use.
The text was updated successfully, but these errors were encountered:
This is a patent "temporary" fix to prevent extremely large models, such as cascade models, from being forced to fully load. I will track the resolution of this in an issue: #312
This is a patent "temporary" fix to prevent extremely large models, such as cascade models, from being forced to fully load. I will track the resolution of this in an issue: #312
This is more pertinent to the worker use case but certain recent changes by ComfyUI and myself have highlighted that there could be some better finessing of memory conditions. While I do feel that it should be largely in the realm of ComfyUI internals to manage this, that project has changed or added a large amount of functionality which is harmful to the worker use case.
The text was updated successfully, but these errors were encountered: