Deep Learning graphic card requirement

5048
18
Jump to solution
10-18-2019 10:12 AM
VHolubec
Esri Regular Contributor

Hello,

I would like to ask, what is the minimal requirement to graphic card for deep learning in Esri (Python API learn module / ArcGIS PRO - Deep learning tools). 

I have installed as documented - fastai=1.0.39 pytorch=1.0.0 torchvision, and it seems that everything works well. The probelm begins with data.show_batch() where I get error - Found GPU0 Quadro K4000 which is of cuda capability 3.0

It seems, that this GPU is no longer supported by underlying PyTorch.

I have 2 questions: 

1) would it be possible to document it somewhere - which cards are supported and which does not

2) Is there any workaround how to run Deep learning on machine with Quadro K4000?

Thank you.

Vladimir

Vladimir
0 Kudos
18 Replies
DanPatterson_Retired
MVP Emeritus

from the help, if you downloaded it, in...

GPU processing with Spatial Analyst

Currently, only NVIDIA GPUs with CUDA compute capability of version 3.0 or later are supported.

There are more topics and discussion with a 'cuda' search in the help files

0 Kudos
VHolubec
Esri Regular Contributor

Dear Dan,

It CC 3.0 would be supported by PyTorch - the NVIDIA CUUDA K4000 should works fine by - CUDA - Wikipedia. But as Shaun Walbridge wrote CC 3.5 is needed and there K4000 is really not supported.

Vladimir
0 Kudos
ShaunWalbridge
Esri Regular Contributor

OK, I expect some of these issues to be worked out before the release goes final, and for the documentation to be clarified around what is happening. If you do have specific issues with Python packages in the beta, please feel free to use the EAP forum to post details and we can take a look.

Cheers,

Shaun

0 Kudos
VHolubec
Esri Regular Contributor

Dear Shaun,

this issue I met in Jupyter from PRO Final 2.4. I don´t know now, what Learn Module version was there. Testing on 2.5 is still on my To-Do list.

Vladimir
0 Kudos
ShaunWalbridge
Esri Regular Contributor

OK, yes if you could test on 2.5 that's helpful since the Python API version included (1.7.0) has a significantly different arcgis.learn module. If you think the error handling for environments with insufficient CUDA compute versions should be improved, I recommend adding an issue to the Python API repository: GitHub - Esri/arcgis-python-api: Documentation and samples for ArcGIS API for Python 

Cheers,

Shaun

VHolubec
Esri Regular Contributor

Dear Shaun, ok, thank you. I will duplicate this issue to the github. Thank you for help.

Vladimir
0 Kudos
DanPatterson_Retired
MVP Emeritus

Robert

PyTorch 

supports windows

0 Kudos
DanPatterson_Retired
MVP Emeritus

Kory Kramer

Can you pass this on the appropriate team if it isn't in the Imagery and Remote Sensing‌ realm

KoryKramer
Esri Community Moderator

For questions/problems with PyTorch installation, maybe Shaun Walbridge‌ could help more than somebody in the imagery space...