Skip to content
This repository was archived by the owner on Jan 12, 2026. It is now read-only.

Add CUDA kernels for grid sampling#31

Merged
CarloLucibello merged 7 commits intoFluxML:masterfrom
pxl-th:master
Nov 15, 2021
Merged

Add CUDA kernels for grid sampling#31
CarloLucibello merged 7 commits intoFluxML:masterfrom
pxl-th:master

Conversation

@pxl-th
Copy link
Member

@pxl-th pxl-th commented Nov 2, 2021

GPU part for the NNlib grid sampling PR.

@DhairyaLGandhi
Copy link
Member

Very cool! Thanks! Could we perhaps make the regular sampling code work with CuArrays directly?

@pxl-th
Copy link
Member Author

pxl-th commented Nov 2, 2021

The code is now re-using almost everything from the NNlib, except for the _safe_add! function which needs to do atomic addition and CUDA kernel indexing stuff.

@pxl-th pxl-th marked this pull request as ready for review November 11, 2021 14:48
@CarloLucibello CarloLucibello merged commit 7710e7d into FluxML:master Nov 15, 2021
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants