lllyasviel / lllyasviel/FramePack
RTX 5090 - Extremely low speed, 2-3x slower than Illyasviel's tests - 5.04-5.6/s Unoptimized
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 17.3k
- Forks
- 1.7k
- PR merge metrics
- No merged PRs in 30d
Description
Running into this extremely bizarre behavior. By Illyasviel's tests, I should be getting even better scores than them.
I have done the troubleshooting steps recommended [here](https://github.com/lllyasviel/FramePack/issues/151#issuecomment-2817054649).
I have 64GB of 6800MT/s ram, and FramePack is installed on a 2 year old NVME 1.4 SSD. It's in good health and has more than 100GB of space.
My system drive is also a Samsung NVME with over 100GB of space.
Of note is that due to having a 50 series card, I had to set up a custom conda environment with a cuda 12.8 install and Torch 2.7.1+cu128. I don't see how this would reduce the speed though, if anything, the optimizations to torch should improve it.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No source file, test, or entry point is named. Start by reproducing the reported RTX 5090 performance with the custom conda environment, CUDA 12.8, and Torch 2.7.1+cu128, then compare it with Illyasviel's reported results; done means identifying and resolving the cause of the 2–3x slowdown.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning, performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100