How do I get past 'GPU not available on #Azure?'

I can create a virtual machine using my subscription… but a VM instance with a powerful NC series GPU gives me an error of NotAvailableForSubscription. What do I do??

Transcribed from source info from Azure Solutions Architects, with our thanks:

Please note: The NCasT4v3, ND A100 v4, NDM A100 v4, and NDv2 are constrained globally, and we are not able to support sponsorship requests for these SKUs for the foreseeable future.

Please also keep in mind that due to constraints, the number of GPUs and the region you select may need to be adjusted depending on the demand in the region.

In some instances, choosing a different region, VM Family, and/or different quantity of GPUs may yield a different result.

With these factors in mind, this is the process to request access to GPU’s when needed:

GPU Request Process

  • Determine the Azure Service and VM Family size you are looking for and calculate the total CPU quantity needed.

    • Example: A Standard_NC6s_v3 has 1 GPU but 6 CPU, so you would request 6 CPU)
  • From within the subscription, start the CPU Quota increase request process and ensure the appropriate Azure service is specified.

    • Examples: Virtual Machine, Azure Machine Learning, Azure Lab Services
  • GPU optimized virtual machine sizes

  • Request a CPU Quota increase

  • Work with Support on the request.

  • If Support denies the request, e-mail AzureUGrants@microsoft.com and they will escalate your Support Incident number to our internal contact and work with Support on getting the request approved. (You might mention CloudBank.)

  • Requests that get approved are typically fulfilled within 48 hours of escalation.

NOTE!! Owing to an existing component shortage combined with pandemic impact on supply chains, there is a shortage of GPUs. This may be reflected in your cloud account as a lack of permission to use certain GPU or instance types. There are currently two approaches to this “road block”. First: Use the built-in quota management tools for that cloud to request access. Second: Look for alternative GPU types that may be less state-of-the-art but can still help you get the job done.