This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](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(19.2, 25%, 11%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EAS%3C/text%3E%3C/svg%3E)
Subscription ID : e30f2093-840c-43cf-8aa6-35923b617406
I have created a Machine Learning resource and then uploaded a Jupyter Notebook that I have previously runned in Google Colab.
In the Google Colab free GPU my Deep Learning model for Object Localization (Yolov5) runs in a heavier model (Yolov5l.pt for 100 epochs) in 4.5 hours and in Azure my model in its slightest version (Yolov5s.pt) runs for 12 hours and it is only in the half of the process (41 epochs out of 100).
Why is it so slow ?
What can I change to make it faster ?
FYI: In the second image I show you that I have uploaded in this place my training dataset.
![](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(22.400000000000002, 11%, 12%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ERM%3C/text%3E%3C/svg%3E)
@Andreas Stathopoulos Thanks for the question. We have forwarded to the product team to check on this.
@Andreas Stathopoulos Can you please confirm which GPU you got in Google Collab?. They are known to hand out P100s initially and then go to K80s, whereas here you are using a K80.
![](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(92.8, 74%, 18%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EA%3C/text%3E%3C/svg%3E)
Same problem for me!
Though I have in Jupiter:
Num GPUs Available: 1
My tensorflow models are calculating slower than on my local Core i5 PC without any GPU.
![](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(144, 59%, 23%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EAR%3C/text%3E%3C/svg%3E)
Have you tried looking at the server IOPS? NC6 Promo only supports Standard disks which have an IOPS ceiling of 500.