How to test a plugin quickly

Osama
Level 1
How to test a plugin quickly

I installed the deep learning GPU plug-in

It requires specific versions of CUDA and cudNN. I have different versions.

It would be very helpful if there was a dead-or-alive for the plugin that allowed me to check quickly if the plugin actually works or not.

Incidentally, there is a misspelling on the usage page of this plugin, currently says

"Here you can compute the instanciations of the components of this plugin across projects."

In English, should have been instantiations

Thanks

Osama Neiroukh

 

0 Kudos
2 Replies
CoreyS
Dataiker Alumni

Hey @Osama thanks for pointing out and welcome to the Dataiku Community. While you wait for a more detailed response, please see this previous thread: deeplearning plugin to work with cuda 9.0 

Just to confirm which page did you see the error on? For example I did not find it on the plugin page on the Dataiku website: https://www.dataiku.com/product/plugins/deeplearning-image-cpu/ 

Looking for more resources to help you use Dataiku effectively and upskill your knowledge? Check out these great resources: Dataiku Academy | Documentation | Knowledge Base

A reply answered your question? Mark as โ€˜Accepted Solutionโ€™ to help others like you!
0 Kudos
Osama
Level 1
Author

Thanks Corey for quick response.

The spelling error is on the *GPU* version of the plugin, not the CPU one.

I realize the NVidia stack of drivers and libraries can be challenging to setup and don't expect Dataiku to take provide any directions for it. I was simply looking for a way to verify if the GPU plugin works without having to back down my CUDA/cudnn/driver combo which takes a lot of delicate effort to get working in unison. Since there is some dependency on the driver, creating isolated environments or containers is not an option.

I looked at the link you sent me. I'm actually on CUDA 11.1 and cudnn 8.05 as I wish to use TensorFlow 2.X on the side. Anyway the page gave me some tips on how I might progress.

 

 

0 Kudos