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.

Sign up for GitHub

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

Jump to bottom

Integration of oneDNN for CPU operations #1013

Open
1 of 13 tasks
avik-pal opened this issue Sep 2, 2024 · 1 comment
Open
1 of 13 tasks

Integration of oneDNN for CPU operations #1013

avik-pal opened this issue Sep 2, 2024 · 1 comment
Labels
luxlib speculative

Comments

@avik-pal
Copy link
Member

avik-pal commented Sep 2, 2024 •
edited
Loading

  • Add oneDNN binaries to Yggdrasil -- feat: add oneDNN CPU build JuliaPackaging/Yggdrasil#9345
  • how expensive is it to construct the internal memory type for the operations?
  • create a oneDNN.jl lib that is framework agnostic. High level functionality along with integration with Enzyme and ChainRulesCore
    • convolution
    • batchnorm
    • groupnorm
    • fused_dense (low-priority)
  • integrate the primitives into LuxLib (via extensions)
    • convolution (highest priority)
    • activations (low-priority)
    • batchnorm
    • groupnorm
    • fused_dense (low-priority)
@avik-pal
Copy link
Member Author

avik-pal commented Nov 3, 2024

This is mostly speculative at this point. With Reactant this wouldn't be needed mostly

@avik-pal avik-pal transferred this issue from LuxDL/LuxLib.jl Nov 3, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
luxlib speculative
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@avik-pal