fastai 使用谷歌云平台

Question

for studying fastai course I am trying to set up Google Cloud Platform (GCP) to use pyTorch and fastai using fastai guide.

I created a google project and want to create an instance using machine type "n2d-highmem-8'" in zone "us-west1-b" as I suppose to do by the guide.

I get err: "Machine type with name 'n2d-highmem-8' does not exist in zone 'us-west1-b'."

• when I looked for other zones that contain machines with the type 'n2d', I found "europe-west4-b", but when I try to run it with this zone I get the err: "ERROR: (gcloud.compute.instances.create) Could not fetch resource:
  • The resource 'projects/.../zones/europe-west4-b/acceleratorTypes/nvidia-tesla-p100' was not found"

there some things I can not understand here.

• by creating a project I set its zone? I thought project is like a class and I can set different zones for different instances.

• I can not understand why I can not run it on 'us-west1-b', I trien to change to 'n2d-highmem-4' and 'n2d-highmem-16' but it did not work.

I created the instance with google cli.

does quota mean the number of instances I can create of a project? is there any reason I will need more than 1?

---

first of all, thank you..

I still can not make it works, and I fail to understand the basics here. now it seems like its start to creating the instance but now I get the error: WARNING: Some requests generated warnings: - Disk size: '200 GB' is larger than image size: '50 GB'. You might need to resize the root repartition manually if the operating system does not support automatic resizing the disk means the google cloud instance disk size right? its the vm not my own computer disk memory right? why is it a problem that the vm size is larger the image.

I thought it has to be like it.

Answer 1

You may want to try n1-highmen-8 instead of n2d-highmem-8, that's from the Fastai Forum post by Satyartha(April 19)

Answer 2

according to this article n2d machien type is not available in US-west regions adn zones.

For the Disksize error, it seems to be FastAI related .