Google Cloud Platform quotas not always displayed

Question

I use GCP with several identical projects. For each new project, I need quotas for one GPU (Tesla K80). To apply for increasing my GPU quota, I open the console and go to "IAM and Admin"> "Quotas", There I filter for my region (europe-west-1) and look for the entry "NVidia K80 GPU".

I noticed that the Compute Engine APIs appear only after visiting the Compute Engine menu at least once. So far, so good. However, the option for GPUs appears only after repeatedly viewing and switching between projects and revising the quota page. It seems completely random.

Here is an example of two identical projects and available quota options:

The project " examplestudent02 " has a GPU option:

The project " examplestudent03 " has no GPU option:

I can not understand what this parameter does. Has anyone experienced something similar? Is there anything that needs to be activated before the GPU quota option appears?

There is a related question about stack overflow. However, the GPU parameter also does not appear when changing the Quota type to "All quotas" (which is the default by default). Also, switching to the quota page in incognito mode did not help. Finally, I usually use Chrome, but I also tried logging in with a different browser (Firefox), which also didn't help.

Answer 1

The following answer is based on the current scenario, since there are no more problems with displaying quotas.

To answer your question, you can go to the quota page of your console and check your GPU quota. Currently, GPUs are listed under the name of NVIDIA K80 GPUs and NVIDIA P100 GPUs (not just GPUs). They can be easily filtered by selecting from the Metric column on this page.

If none of them can be found, a quota may not have been assigned. To request a GPU quota, you can follow the steps mentioned in this article . After submitting the request, it may take 24 to 48 hours to get approval.

In doing so, you need to keep in mind that free trial accounts do not get the default GPU quota . I would also suggest checking limitations on instances using the GPU to avoid any future issues.

Hope this helps.