I’ve used colab all the time and but just want to set my own jupyter config, I determined to use GCP.
Following the fast.ai gcp guide, waiting for acceptance of quota,
I settled the area of Compute Region and Zone as asia-east. (I’m living in South Korea)
But as for fast.ai’s manual the region is settled to us-west1-b and no other comment of option.
I searched the document and they say it related to price, round-trip time… best-practices-compute-engine-region-selection
As far as I know, after I initiate instance, I cannot change the region.
Hi Dionne, I live in Hong Kong, and wonder whether I should select some Asia region too. Did you find out the answer to your question? btw, believe you can still change the region later. The guide says "you can modify this later with gcloud config set compute/zone NAME"
To tell you my answer first, I concluded colab is the best option.
Yes, you should select depends on your region. Theoretically your region and mine have cloud according to this document: Google Cloud: Region and Zones. But since Tesla P100 (which fits the fastai courses) is available only in Taiwan at Asia, we have no other choice.
And actually that time I succeeded to initiate the instance, installed all the requirements and ran some codes, but the problem was I couldn’t run it again after I stop it It said the resource was not enough.
And for caution’s notice I also tried the AWS, and it spent me so much money.
I can’t sure your case since it’s been 3 months and somewhat living different region,
but I highly recommend you just use colab pro and they give you T4 and P100 $9.99 for a month.
I tried Collab before. The setup looks nice and clean. However, their interface is somewhat different from the traditional Jupyter Notebook interface, which I was not quite used to.
I just set up Google Cloud Platform using Taiwan machines and Tesla P100. It works fine so far. I did have an issue with resources during my setup. In particular, it complained that Quota 'GPUS_ALL_REGIONS' exceeded. The issue was resolved by following the steps on FastAI page and adjusting the GPU quote.
Haven’t been able to get a quota incresase:
"Hello,
We have received your quota request for steadyeye.
Unfortunately, we are unable to grant you additional quota at this time. If this is a new project please wait 48h until you resubmit the request or until your Billing account has additional history.
Your Sales Rep is a good Escalation Path for these requests, and we highly recommend you to reach out to them. "
It said the resource was not enough.
