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

torch.nn.DataParallel(net).to(dml) raised an error #609

Closed
etoyz opened this issue Jul 13, 2024 · 1 comment
Closed

torch.nn.DataParallel(net).to(dml) raised an error #609

etoyz opened this issue Jul 13, 2024 · 1 comment

Comments

@etoyz
Copy link

etoyz commented Jul 13, 2024

Reproduce:

  1. When the second parameter of torch.nn.DataParallel() is None,

  2. Then torch._utils._get_all_device_indices() will be called.

  3. Then lambda m: list(range(m.device_count())) will be called, and it can be proved that the m is <class 'torch_directml.PrivateUse1Module'>

  4. torch_directml.PrivateUse1Module.device_count() seems not to be implemented ???

@joshjkim
Copy link

@etoyz Our latest torch_directml 0.2.4.dev240815 release adds support for device_count with PrivateUse1. Please run pip install torch-directml --upgrade to update to the latest build

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants