•   6 months ago

How to use GPU in Jupyter Notebook?

I have successfully migrated the code from Jupyter Notebook to Azure Jupyter Notebook.
My problem is I can not find a selection to enable GPU for my training. The CPU training is 10x slower than GPU. What is the solution?

  • 8 comments

  • Manager   •   6 months ago

    Hi Chang!

    You can also head over to the hackathon's Slack to ask the Azure Experts your technical questions in #office-hours channel. In the meantime, I'll also try to get answer for you.

    Thanks,
    Khimee

  • Manager   •   6 months ago

    Hi Chang, you can easily enable GPU training by using the Azure Machine Learning SDK inside Azure Notebooks. Here's is some documentation on how to do it: https://docs.microsoft.com/en-us/azure/notebooks/use-machine-learning-services-jupyter-notebooks

    Hope this helps!
    Max

  •   •   6 months ago

    I use Python3.6 - AzureML in Notebook VM. I can select a VM with 4 core CPUs. It worked fine. But I can not create a VM with 6 core CPU + 1 GPU. Looks like the max allocation of a free account is 4 cores.

    Here is the Create Failed error message;
    Operation failed
    Failed to create or update notebook instance compute. Error details: RequestId: 7f490504-2690-420b-9d4e-95f4c68ac092 Error code: 'InvalidTemplateDeployment'. Target: ''. Message: 'The template deployment 'ca5e444d-5357-4d1d-8d5f-861f1648146c' is not valid according to the validation procedure. The tracking id is '7f490504-2690-420b-9d4e-95f4c68ac092'. See inner errors for details. Please see https://aka.ms/arm-deploy for usage details.' Error code: 'QuotaExceeded'. Target: ''. Message: 'Operation results in exceeding quota limits of Core. Maximum allowed: 4, Current in use: 4, Additional requested: 4. Please read more about quota increase at https://aka.ms/ProdportalCRP/?#create/Microsoft.Support/Parameters/{"subId":"1e39e517-b744-4a78-9221-3137676890db","pesId":"15621","supportTopicId":"32447243"}.' .

  • Manager   •   6 months ago

    Hi Chang, yes you are correct. Unfortunately the free account does limit to 4 cores. If you are a student, you can sign up for the Student free account offer which does not limit to 4 cores. https://azure.microsoft.com/en-us/free/students/

    I apologize for the inconvenience,
    Max

  •   •   6 months ago

    Hi can I withdraw my registration? Please let me know soonest.

  • Manager   •   6 months ago

    Hi Annie,

    If you'd like to unregister for the hackathon, you can do so by clicking "unregister" on the right sidebar on the home tab. However, if you'd like to continue and if there's anything we can do to help with the hackathon, please let us know!

    Thanks,
    Khimee

  •   •   6 months ago

    Thank you for reply. I really appreciate that. I managed to opt out from this competition.

  •   •   about 2 months ago

    Hy am working on Recomendation engines but the problem is when i am uploading the more than 10MB dataset in my project its not uploading how can i upload the files more than 10MB?

Comments are closed.