Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

pytorch-directml next release #511

Open
ChiaYen-Kan opened this issue Oct 1, 2023 · 7 comments
Open

pytorch-directml next release #511

ChiaYen-Kan opened this issue Oct 1, 2023 · 7 comments
Labels
pytorch-directml Issues in PyTorch when using its DirectML backend

Comments

@ChiaYen-Kan
Copy link

Will there be a next release version of pytorch-directml?

@AGenchev
Copy link

AGenchev commented Oct 2, 2023

why would it be necessary ? running strange direct-x only GPU ?

@ChiaYen-Kan
Copy link
Author

why would it be necessary ? running strange direct-x only GPU ?

why not???

stable diffusion can use pytorch-directml as backend and i am using it

@wchao1115 wchao1115 added the pytorch-directml Issues in PyTorch when using its DirectML backend label Oct 2, 2023
@Adele101
Copy link

Adele101 commented Oct 4, 2023

Hi @ChiaYen-Kan, thank you for submitting this request. While I can't provide a roadmap as the moment, please know that your feedback is valuable to us. We will follow up once we can review this request.

@ChiaYen-Kan
Copy link
Author

ChiaYen-Kan commented Oct 6, 2023

Hi @Adele101 thank you for reply

I hope in the next release the team of directml can improve pytorch-directml memory usage efficiency
in current release the memory usage of pytorch-directml is poor then other backend, some function in stable diffusion can't work proper

vladmandic/automatic#2250 (comment)
pkuliyi2015/multidiffusion-upscaler-for-automatic1111#93

@shott92
Copy link

shott92 commented Nov 4, 2023

Also curious as to the progress specifically towards python 3.11 compatibility, as per the python.org release notes, it is on average 25% faster than previous versions, and one of the most significant performance improvements in many generations.
It would be amazing to be able to leverage this performance enhancement.

It's notable that the pytorch 2.1 version is more optimised to run 3.11 as such an update to support these components would be a huge boon to the performance of the DirectML platform

@sorasoras
Copy link

Looking forward to support for pytorch2.1 as well

@Dwanvea
Copy link

Dwanvea commented Dec 27, 2023

I hope PyTorch 2.1 will be supported. It would be great for many AMD users to have a better performance in stable diffusion.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
pytorch-directml Issues in PyTorch when using its DirectML backend
Projects
None yet
Development

No branches or pull requests

7 participants