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

Tensor encoding assumptions in advanced path #1925

Open
victor-eds opened this issue Aug 19, 2024 · 0 comments
Open

Tensor encoding assumptions in advanced path #1925

victor-eds opened this issue Aug 19, 2024 · 0 comments

Comments

@victor-eds
Copy link
Contributor

Advanced path introduces assumptions in tensor encoding. Layouts are a powerful mechanism in Triton to implement some algorithms. Also, upstream Triton has extense code to handle layout conversions in an optimized way that we could use (https://github.com/intel/intel-xpu-backend-for-triton/blob/llvm-target/lib/Conversion/TritonGPUToLLVM/ConvertLayoutOpToLLVM.cpp). We should be aware of this when bringing all of the nice features the advanced path brings so we have access to the best of both worlds when writing future optimizations.

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

No branches or pull requests

2 participants