lllyasviel / lllyasviel/stable-diffusion-webui-forge
Forge keeps using more and more RAM after api generations.
- Dominant language
- Python
- Stars
- 13k
- Forks
- 1.7k
- PR merge metrics
- No merged PRs in 30d
Description
Hi, I now use the webui forge for some iterative generating purposes.
I have a lot of model loading/unloading tasks in forge, including LoRAs and ControlNets.
The thing is, whenever I keep going through the generation process, the RAM usage keeps increasing, as if the cached model is not being purged in RAM. It discretely spikes especially when the ControlNet model is loaded. (I use different ControlNet models for iteration)
I now use 4 4090 GPUs. One generation process uses approx 15000MB in a single GPU.
My memory starts in 73GB, ending up overflowing 188GB, which is the device limit.
Any help? Thanks.
Contributor guide
No contributing guide indexed for this repository
Research direction
No files, tests, or entry points are named. Start by reproducing repeated generations while switching LoRAs and ControlNets, and monitor RAM from the reported 73GB baseline; done means identifying the retention point and confirming memory remains bounded after iterative generation.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100