Skip to content
This repository has been archived by the owner on Sep 25, 2023. It is now read-only.

[FEA] Reevaluate Windows module for performance improvements #233

Closed
mnicely opened this issue Sep 20, 2020 · 2 comments
Closed

[FEA] Reevaluate Windows module for performance improvements #233

mnicely opened this issue Sep 20, 2020 · 2 comments
Labels
feature request New feature or request

Comments

@mnicely
Copy link
Contributor

mnicely commented Sep 20, 2020

The windows module need to be reevaluated to find areas of further performance improvements.

@mnicely mnicely added the feature request New feature or request label Sep 20, 2020
@awthomp
Copy link
Member

awthomp commented Sep 24, 2020

Note: This includes similar investigations as explored in #235. Developers should be examining:

  • Use of numpy over cupy for small-scale calculations like overlap size, FFT size, and other generic parameters that don't require GPU compute
  • Elementwise kernels, where appropriate (either raw cupy, as denoted in the above PR, or Numba)
  • Caching of FFT plans

@mnicely
Copy link
Contributor Author

mnicely commented Oct 16, 2020

Closed by #267

@mnicely mnicely closed this as completed Oct 16, 2020
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
feature request New feature or request
Projects
None yet
Development

No branches or pull requests

2 participants