New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#1

Post by jamiis » Fri, Apr 17, 2015, 4:14 PM UTC

New GPU Optimized EC2 G2 Instance, the G2.8xlarge
aws.amazon.com

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#2

Post by moonbug » Fri, Apr 17, 2015, 4:38 PM UTC

4 of the current Grid K520 GPUs in a single instance, at 4x the price (on-demand and spot) of the of the g2.2xlarge. Shame they're not also rolling out a Maxwell hardware-backed instance type.

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#3

Post by michaelt » Fri, Apr 17, 2015, 4:46 PM UTC

It would be nice if there was an AMI for the latest LTS version of Ubuntu with the CUDA drivers installed and working.

Last time I tried some CUDA work on AWS I spent the best part of a day trying to just get the drivers set up right for a basic test app to work on 14.04 - eventually I gave up and went back to 12.04 which, in 2015, didn't give me much confidence in CUDA as something I can expect decent support for.

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#4

Post by alepper » Fri, Apr 17, 2015, 4:57 PM UTC

michaelt wrote: post #3

It would be nice if there was an AMI for the latest LTS version of Ubuntu with the CUDA drivers installed and working. Last time I tried some CUDA work on AWS I spent the best part of a day trying to just get the drivers set up right for a basic test app to work on 14.04 - eventually I gave up and went back to 12.04 which, in 2015, didn't give me much confidence in CUDA as something I can expect decent support for.

I have this kicking around from some experiments a month or two ago:

    sudo apt-get update
    sudo apt-get install build-essential linux-image-extra-virtual
    sudo reboot
    echo options nouveau modeset=0 | sudo tee -a /etc/modprobe.d/nouveau-kms.conf
    cat > /etc/modprobe.d/blacklist-nouveau.conf
    blacklist nouveau
    blacklist lbm-nouveau
    options nouveau modeset=0
    alias nouveau off
    alias lbm-nouveau off
    
    sudo update-initramfs -u
    sudo reboot
    sudo apt-get install linux-headers-$(uname -r)
    wget http://uk.download.nvidia.com/XFree86/Linux-x86_64/346.35/NVIDIA-Linux-x86_64-346.35.run
    chmod +x NVIDIA-Linux-x86_64-346.35.run
    sudo ./NVIDIA-Linux-x86_64-346.35.run
    nvidia-smi -q | less

This is obviously not production-ready, and is heavily cribbed from online (I couldn't quickly re-find where) but is good enough if you just want to play.

EDIT: I think this was the original: http://ubuntuhandbook.org/index.php/2015/01/install-nvidia-3...

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#5

Post by moonbug » Fri, Apr 17, 2015, 4:59 PM UTC

michaelt wrote: post #3

It would be nice if there was an AMI for the latest LTS version of Ubuntu with the CUDA drivers installed and working. Last time I tried some CUDA work on AWS I spent the best part of a day trying to just get the drivers set up right for a basic test app to work on 14.04 - eventually I gave up and went back to 12.04 which, in 2015, didn't give me much confidence in CUDA as something I can expect decent support for.

Canonical's packaging of the Nvidia drivers is risible. It really is easier just to pull the driver and CUDA installers from the nv site and install manually. Or use the Amazon Linux AMIs which just DTRT.

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#6

Post by Sanddancer » Fri, Apr 17, 2015, 5:02 PM UTC

moonbug wrote: post #2

4 of the current Grid K520 GPUs in a single instance, at 4x the price (on-demand and spot) of the of the g2.2xlarge. Shame they're not also rolling out a Maxwell hardware-backed instance type.

They're not even Grid K520s, they're K5000s. They've only got 4 gigs of RAM each, instead of the 5 gigs for teslas or 8 gigs for the K520s. This sounds a lot like they're trying to pull a little bit more value out of their current gpgpu rigs as they slide further and further into obsolescence.

Edit:

Looking again at the pricing and specs, it looks like they're just letting you have a full gpgpu box, instead of the multi-tenant box they previously offered where you share with three other people.

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#7

Post by moonbug » Fri, Apr 17, 2015, 5:16 PM UTC

Sanddancer wrote: post #6

moonbug wrote: post #2

4 of the current Grid K520 GPUs in a single instance, at 4x the price (on-demand and spot) of the of the g2.2xlarge. Shame they're not also rolling out a Maxwell hardware-backed instance type.

They're not even Grid K520s, they're K5000s. They've only got 4 gigs of RAM each, instead of the 5 gigs for teslas or 8 gigs for the K520s. This sounds a lot like they're trying to pull a little bit more value out of their current gpgpu rigs as they slide further and further into obsolescence. Edit: Looking again at the pricing and specs, it looks like they're just letting you have a full gpgpu box, instead of the mu…

Well, the K520 is a single AIB with two discrete GPUs on it, each being a GK104 with 4GB (which seems to match the spec of the K5000). A g2.2x instance has a single GPU - ie half a K520 - assigned to it.

Your claim surprised me enough to stand an instance up and take a look for myself. Sure enough, they report having 4 K520 devices.

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#8

Post by sparaker » Fri, Apr 17, 2015, 5:18 PM UTC

This is very good addition. Thanks a lot.

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#9

Post by msoad » Fri, Apr 17, 2015, 5:33 PM UTC

michaelt wrote: post #3

It would be nice if there was an AMI for the latest LTS version of Ubuntu with the CUDA drivers installed and working. Last time I tried some CUDA work on AWS I spent the best part of a day trying to just get the drivers set up right for a basic test app to work on 14.04 - eventually I gave up and went back to 12.04 which, in 2015, didn't give me much confidence in CUDA as something I can expect decent support for.

Yes, considering how expensive(in terms of price per hour) it is to just manually install drivers on those machines.

Re: New GPU Optimized EC2 G2 Instance, the G2.8xlarge

#10

Post by Sanddancer » Fri, Apr 17, 2015, 6:32 PM UTC

moonbug wrote: post #7

Sanddancer wrote: post #6

Earlier quoted context omitted.

They're not even Grid K520s, they're K5000s. They've only got 4 gigs of RAM each, instead of the 5 gigs for teslas or 8 gigs for the K520s. This sounds a lot like they're trying to pull a little bit more value out of their current gpgpu rigs as they slide further and further into obsolescence. Edit: Looking again at the pricing and specs, it looks like they're just letting you have a full gpgpu box, instead of the mu…

Well, the K520 is a single AIB with two discrete GPUs on it, each being a GK104 with 4GB (which seems to match the spec of the K5000). A g2.2x instance has a single GPU - ie half a K520 - assigned to it. Your claim surprised me enough to stand an instance up and take a look for myself. Sure enough, they report having 4 K520 devices.

Gotcha. I was looking at the spec sheets of the various instances, and could only find that they are GK104s. I didn't think to look to see if there was a dual GK104 board, because video card makers tend to do that only for their high end boards. Definitely makes sense, though, for the density Amazon would want to run in.