Azure Machine Learning
An Azure machine learning service for building and deploying models.
2,887 questions
Install Nvidia Drivers on azure ml compute instance
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(204.8, 4%, 29%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ECA%3C/text%3E%3C/svg%3E)
Christos Athanasiadis
0
Reputation points
I am new to Azure world and have created a workspace and a machine learning compute instance with an A100 GPU. I have also added a Cuda-base environment to this compute instance. Moreover, I am trying to install PyTorch and Nvidia drivers to run my ML models using GPU support.
However, I am lost on properly installing the Nvidia drivers in my compute instance. I tried to follow these instructions here, but that didn't work for me. When I run the following command:
nvidia-smi
I am receiving the following error:
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running
Is there any suggestion on properly installing the Nvidia drivers in my compute instance?
{count} votes
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(6.4, 61%, 10%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ES%3C/text%3E%3C/svg%3E)
santoshkc 8,690 Reputation points Microsoft Vendor2024-09-27T08:23:06.3266667+00:00 Thank you for reaching out to Microsoft Q&A forum!
It looks like the NVIDIA drivers are not installed configured correctly on your Azure Machine Learning compute instance. Since you're using a GPU-powered compute instance (with an A100 GPU), Azure typically installs the required NVIDIA drivers automatically, but there can be issues with driver installation or environment setup.
Please look into the documentation to install the NVIDIA drivers:
I hope this helps. Thank you.
-
Christos Athanasiadis 0 Reputation points
2024-09-27T09:04:37.0766667+00:00 Hey, the first link is about NVIDIA GPU Driver Extension for Windows on a VM. In my case, the OS of my compute instance is Ubuntu and the instructions between a typical VM with a compute instance seem to be different. The same goes with the second link, it seems that you provided instructions on how to install NVIDIA drivers on Windows :(.
It looks like the NVIDIA drivers are not installed and configured correctly on your Azure Machine Learning compute instance: Exactly, I thought the same that the drivers would be automatically installed after using also the proper environment!
-
santoshkc 8,690 Reputation points Microsoft Vendor
2024-09-30T13:32:51.92+00:00 Thank you for your follow-up query.
To utilize GPU support on your Azure Machine Learning compute instance with an A100 GPU, ensure your environment is properly configured. Use Ubuntu's
ubuntu-driversutility to install NVIDIA proprietary drivers, as these are specifically packaged for compatibility and security. After installation, reboot the instance to finalize the setup. Once the drivers are correctly configured, you can install libraries like PyTorch to leverage GPU acceleration for your machine learning models.
See: Ubuntu packages NVIDIA.I hope this helps. Thank you.
-
Christos Athanasiadis 0 Reputation points
2024-09-30T14:22:22.26+00:00 After checking a bit further, I concluded that the GPU might not be working for me. When I am trying to identify which GPU I am running to using the following command from the terminal:
sudo lshw -C displayThe result is as follows:
*-displaydescription: VGA compatible controllerproduct: Hyper-V virtual VGAvendor: Microsoft CorporationSomehow it seems that while I have added an A100 when trying to identify it in my compute instance I cannot locate it!
-
santoshkc 8,690 Reputation points Microsoft Vendor
2024-10-01T08:22:01.1233333+00:00 Sorry for the trouble you are facing. I recommend reporting this issue to the Azure Machine Learning support team. They will be able to investigate the issue further and provide a more targeted solution. You can report the issue by following these steps:
- Go to the Azure portal and navigate to your Machine Learning Services resource.
- Click on the "Support + troubleshooting" tab.
- Fill out the required information, including a detailed description of the issue and any steps you have taken to troubleshoot it.
- Submit the support request.
The Azure Machine Learning Services support team will review your request and provide assistance as soon as possible.
Sign in to comment
Sign in to answer