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.

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

Update the recipe url to the new format #55

Open
leofang opened this issue Mar 18, 2023 · 7 comments
Open

Update the recipe url to the new format #55

leofang opened this issue Mar 18, 2023 · 7 comments

Comments

@leofang
Copy link
Member

leofang commented Mar 18, 2023

The recipe should be updated to follow the template form as in cuTENSOR/cuSPARSELt:

This would allow the CF bot to pick up the new updates from NVIDIA and send a version update PR to here automatically.

@leofang
Copy link
Member Author

leofang commented Apr 26, 2023

This is done in #56.

@leofang leofang closed this as completed Apr 26, 2023
@leofang
Copy link
Member Author

leofang commented May 4, 2023

I was wondering why this feedstock remains silent since #56... There are a few newer versions already, but the bot didn't send any update.

It turns out the cuDNN team uses a wrong format in the json file (the package build number is mistakenly included in the json filename) and we have no way to check it. It was already raised by another internal team: nvbugs 4056484

@leofang leofang reopened this May 4, 2023
@leofang
Copy link
Member Author

leofang commented May 4, 2023

By json files I refer to those under https://developer.download.nvidia.com/compute/cudnn/redist/.

@jakirkham
Copy link
Member

Is it worth raising as an upstream issue?

@jakirkham
Copy link
Member

jakirkham commented Feb 13, 2024

Tried requesting a bot update ( #74 ) to see what would happen, it opened PR ( #75 ) and promptly failed. Details about the failure in this comment: #75 (comment)

Think cuda_major is confusing for the bot. Though there may not be enough information for the bot to go off of either

{% set cuda_major = (cuda_compiler_version|default("12.0.0")).split(".")[0] %}

Edit: The bot is also pinning to CUDA 11.0 ( #75 (comment) ), which causes issues

@jakirkham
Copy link
Member

Did spot one issue: #76

Fixing with PR: #77

Not sure if that is enough to fix this issue, but we can test after that lands

@jakirkham
Copy link
Member

jakirkham commented Feb 13, 2024

The above PR was merged

Tried a fresh version update: #79

Though am still seeing issues with the update. Looks like it may be a different issue though. It is no longer running into issues with cuda_major, but it still doesn't know how to handle CUDA 11 & 12

Edit: More details in this comment: #79 (comment)

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

No branches or pull requests

2 participants