Skip to content

Conversation

CarloLucibello
Copy link
Member

@CarloLucibello CarloLucibello commented Feb 24, 2022

With JuliaGPU/CUDA.jl#1380 merged in CUDA#master, the possibility to have sparse cuda arrays as graphs' backend and supporting graph algebraic operations on gpu becomes more real.

This PR replaces #66, and adds some workarounds that almost gets us there. Still some tests are failing though.

Related to #134.

Related CUDA issues are
JuliaGPU/CUDA.jl#1402 (fixed)
JuliaGPU/CUDA.jl#1403 (fixed)
JuliaGPU/CUDA.jl#1406
JuliaGPU/CUDA.jl#1407 (fixed)

@CarloLucibello CarloLucibello changed the title Cl/cuda2 sparse cuda arrays support Feb 24, 2022
@zsz00
Copy link

zsz00 commented Mar 31, 2022

What is the pr state now ?

@CarloLucibello
Copy link
Member Author

recent CUDA.jl PRs improving sparse support haven't made a tagged release yet (maybe they will be on CUDA 4.0?).
I'll wait for those changes to be released and then bring this PR to completion since I don't think there are further improvements planned in CUDA.jl for the near future. Sparse cuda support will still have many rough edges but definitely we will see some nice improvements.

@casper2002casper
Copy link

Have these been PR's been merged yet?

@CarloLucibello
Copy link
Member Author

closing in favor of #220

@CarloLucibello CarloLucibello deleted the cl/cuda2 branch July 18, 2024 07:23
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

Successfully merging this pull request may close these issues.

3 participants