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

Magma #1

Open
buybackoff opened this issue Apr 11, 2020 · 0 comments
Open

Magma #1

buybackoff opened this issue Apr 11, 2020 · 0 comments

Comments

@buybackoff
Copy link
Member

https://icl.cs.utk.edu/magma/

This is the only native provider that is interesting to test. It has the most important routines that could probably be seamlessly integrated with some init work + codegen.

For CPU OB and MKL pair is more than enough. Atlas & Eigen are not faster or more difficult to use. AMD's one is no more.

Magma's build instructions are quite complicated for Windows, could not do in 5 mins with llvm + conda stack as in OB instructions. Not tried MinGW though.

PyTorch uses Magma and it works on Windows, but they statically link it pytorch/pytorch#7461.

But there should be a script that builds Magma for PyTorch.

Here is a link for static Magma library https://pytorch.org/docs/stable/notes/windows.html

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

1 participant