-
-
Notifications
You must be signed in to change notification settings - Fork 13
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
Add scatter for CUDA support #1
Conversation
Wait for merge of JuliaGPU/CUDA.jl#842 |
Now this PR relies on NNlib#master (or next patch release of v0.7.19) and CUDA#master (or next patch release of v3.0.4). |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
@maleadt Could you trigger the CI test again? I think it is ready to go. |
You can always amend and force-push to re-trigger CI. |
You can also close and reopen to retrigger |
I need at least NNlib v0.7.19 and CUDA v3.0.4 to pass CI, but buildkite just have the following version. [052768ef] CUDA v3.0.0
[f6369f11] ForwardDiff v0.10.18
[872c559c] NNlib v0.7.18 As I know, there should be compatible with current version. |
Then you need to bump the versions in the Manifest. Buildkite doesn't use specific package versions. |
Thank you. It works. |
The compat bounds should also be updated |
The compat bounds look somewhat like this? |
And this other little piece is done, thank you for all these efforts! |
FWIW, it would have been better to squash and not merge this PR. Now there's 13 commits added to master for a single feature, many of which fail CI (complicating operations like |
Transfer from FluxML/NNlib.jl#296.