lllyasviel / lllyasviel/stable-diffusion-webui-forge

FLUX - NF4 uses a lot of RAM and VRAM when a LORA is being used.

Open
#1,218 23 comments 7 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
13k
Forks
1.7k
PR merge metrics
No merged PRs in 30d

Description

For some reason dev fp8 works better than nf4 when I use lora, Nf4 just use a lot of vram and ram making the generation speed absurdly slow.

With dev fp8 I'm getting speeds like 1.01 s/it and with NF4 i;'m getting like 29s/it.

I think its due to lora being patched and vram/ram not being released after that.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No file, test, or entry point is named. Start by reproducing the reported FLUX comparison between dev fp8 and NF4 with a LoRA, then trace LoRA patching and memory release; done means NF4 no longer retains excessive RAM or VRAM and generation speed is comparable under the reported setup.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.