My request limit is set to 0 for all GPU instances across all regions in AWS

I am unable to create any p2 or p3 instances in any of the regions. The limit is set to 0. Is that the default limit? How can I increase this limit?

Regards
Dipjyoti

Do you have a valid payment instrument (credit card) registered with AWS Billing? Mine had expired and I was not able to create instances as well (I got a different error though). Note that you need a valid payment instrument even if you have AWS Credit applied.

Else you can open a case with AWS Support.

Screenshot from 2017-10-30 10-04-28

Then

Then

Screenshot from 2017-10-30 10-06-03

Thanks @anandsaha . I opened a case regarding the same in the Support.

Btw, I do have valid payment instrument and also can create non-GPU instances. Any inputs regarding that?

No other inputs as such. The limit increase request is what it takes. (see here)

1 Like

Be sure to mention ‘fast.ai’ when you request the limit increase. AWS know to look out for cases mentioning this course.

4 Likes

Hi @jeremy,
concerning the limit increase request, is ONE enough for the course ?
or 2?
Thank you.

Yes one is fine.

BTW I heard from AWS that they will be increasing limits this week for everyone who added their AWS ID to the spreadsheets :slight_smile:

2 Likes

Okay, good to know.
that’s great !!!
Thank you so much

I have raised the limit increase request last week and today I got my limits increased. Maybe they have started increasing the limit of everyone in the spreadsheet :slight_smile:

Yes they’ve done it for 75% of our students now.

should we request? or they will increase the limit for all of the students in this course?

You should request - in theory they’re doing it for everyone, but doesn’t hurt to ask if they’ve not gotten to you yet.

Put in a support request. You should be approved in about 24 hours

Hello,
My request has been rejected, below is their reply:

We’re sorry but we’re unable to process the limit increase you requested for p3.2xlarge at this time. As your account is relatively new with us and there is limited usage so far, we would need you to first utilize your current instance capacity before granting such a request. These limits are put in place to help you gradually ramp up activity and decrease the likelihood of large bills due to sudden, unexpected spikes. Once we have a broader window of usage on your account to review, we are happy to reassess any requests.

@jeremy, please is it possible to get a help on this.

Update: I’ve just seen that I have a limit of 5 instances p3.2xlarge on US east N.Virginia region as
my request was for UE Ireland because I’m in Morocco and that’s the closest with p3 instances. So won’t this be a problem if I use the N. Virginia one considering latency as it is far from here?

I wouldn’t suggest using p3 on the whole for this course. They’re expensive and really only useful if you want to run a huge model for a long time

Ok, thanks a lot.

How long did it take to get a response from AWS? I submitted mine before class (mentioning fast.ai), and hope to get this instance up and running tomorrow …

I seem to be having the same issue. I just sent in a request to amazon for a limit increase for p2.xlarge. Here’s hoping it’s approved shortly.