Error with GPU in deep learnig training

[jesusangel]

I am using arcgis 3.2.2, with the latest deep learning package, I am trying to train a neural network using the dedicated GPU of my system. Windows tells me that my graphics card is GPU1, but if in arcgis I select GPU ID: 1, it sends me the following error:

Traceback (most recent call last):
File "c:\program files\arcgis\pro\Resources\ArcToolbox\toolboxes\Image Analyst Tools.tbx\TrainDeepLearningModel.tool\tool.script.execute.py", line 29, in <module>
torch.cuda.set_device(arcpy.env.gpuId)
File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\fastai\torch_core.py", line 72, in _new_torch_cuda_set_device
_old_torch_cuda_set_device(device)
File "C:\Program Files\ArcGIS\Pro\bin\Python\envs\arcgispro-py3\Lib\site-packages\torch\cuda\__init__.py", line 261, in set_device
torch._C._cuda_setDevice(device)
RuntimeError: CUDA error: invalid device ordinal

Failed script Train Deep Learning Model...
Failed to execute (TrainDeepLearningModel).

[DanPatterson]

ArcGIS Pro 3.2 system requirements—ArcGIS Pro | Documentation

under...

General-purpose computing on a GPU

is your GPU upto date and sufficient?

[CodyPatterson]

Hey @jesusangel 

Long shot, but Python is a zero-indexed language, it may be correct that GPU1 is the first GPU, but in the list, it may actually be GPU 0 when you enter it in. I would attempt to enter 0 in that GPU ID: prompt and see if it changes anything!

Cody

[BobBooth1]

You might verify that your GPU is suitable. See:

https://learn.arcgis.com/en/projects/get-ready-for-deep-learning-in-arcgis-pro/#:~:text=check%20whet...