Sign in
Categories
Migration Mapping Assistant Your Saved List Partners Sell in AWS Marketplace Amazon Web Services Home Help

Deep Learning AMI with Source Code (CUDA 8, Amazon Linux)

Comes with deep learning frameworks configured with CUDA 8. Includes Apache MXNet, Caffe, Caffe2, TensorFlow, PyTorch, Theano, CNTK, Torch and Keras. THIS AMI IS NOT UPDATED ANYMORE. For latest AMI, go to: <a href="https://aws.amazon.com/marketplace/pp/B077GF11NF" target="_blank">https://aws.amazon.com/marketplace/pp/B077GF11NF</a> Release tags/Branches used: MXNet 0.11.0 TensorFlow... See more

Customer Reviews

13
Create Your Own Review

AWS finally delivers a robust Deep Learning AMI that works out of the box

  • By Akshay
  • on 10/03/2017

The previous versions of this AMI required users to follow a readme and manually install CUDA drivers. As a result I was hesitant to use it. However the latest update made barely a month ago, significantly improves the AMI and no longer requires user to install or configure any drivers. Most if not all libraries work out of the box.


Deep Learning AMI is just an Ubuntu image with nothing installed :)

  • By Janosch
  • on 09/27/2017

Two weeks ago I got excited at a presentation from two AWS architects in Cambridge telling us how great it is to use Amazon cloud services because they come with everything pre-installed and you could focus on doing your research.

I did choose the p2.xlarge AMI in Ireland to do deep learning on image data. I tried following several tutorials I found on Amazon websites and smart IT blogs to be successful. The disappointment started with Jupyter not being installed. Then trying to do simple Python import commands resulted in Tensorflow not found, MXnet not found and including other core packages required.
After installing Jupyter and lots of other libraries needed to get tensorflow-gpu support to work it turned out that the Cuda library is not properly installed. See also other reviews.

I have spent now a full day installing stuff from scratch which cost me a lot of time and 10$ in cloud services without much progress. I am not going to give up and will soon be a deep learning professional IT architect. Thank you Amazon for your enticing marketing and despicable software services! Lesson learned :)


Runs Keras under TensorFlow

  • By stang
  • on 04/05/2017

I think most reviews for this AMI are outdated, even the ones from 2017: I've been running Keras under TensorFlow on g2 and p2 instances and it was working in GPU mode. No issues so far.


Easy to use with some limitations

  • By Jim T
  • on 01/18/2017

Easy to set up. Contains most of the software I was looking for: Anaconda, Keras, Theano, TensorFlow, CUDA and cuDNN.

Only complaint is that only the cpu version of TensorFlow is installed.

I tried creating an Conda environment for the gpu version of Tensorflow. Was able to install the TensorFlow-gpu but it seems not able to work with CUDA 7.5.

Tried updating to CUDA 8.0 but so far not very successful.

Other than this one glitch, the AMI works. Keras with Theano/CUDA/cuDNN worked w/o any issue.


Awesome Deep Learning AMI

  • By Fast and Easy
  • on 12/04/2016

Fast and Easy.

Fast: Allows you to detailed hard core deep learning computation really smoothly.

Easy: It was super hard to install all the software on this AMI on my computer (got stuck), but then I found this AMI.

Thanks Amazon!


DOES now have necessary drivers: Most other reviews at this time are out of date

  • By Alex Coventry
  • on 11/27/2016

I initially rejected this AMI on the basis of the current reviews, which state that it doesn't have the necessary NVIDIA drivers. But that is no longer the case. Source (besides verifying myself):

http://www.allthingsdistributed.com/2016/11/mxnet-default-framework-deep-learning-aws.html

"...we recently made a set of tools available to make it as easy as possible to get started: a Deep Learning AMI, which comes pre-installed with the popular open source deep learning frameworks mentioned earlier; GPU-acceleration through CUDA drivers which are already installed, pre-configured, and ready to rock."

The five-star rating might be premature, though... I haven't done more than take it for a spin so far.


Old reviews wrong: everything is pre-installed

  • By AWS_LeoDirac
  • on 11/26/2016

Older reviews of this AMI are out of date. The current version of this AMI comes with GPU drivers, CUDA, and cuDNN all pre-installed, as well as many popular deep learning packages.

I've used a lot of deep learning packages over the years, and spent way too much time installing them and setting them up -- it's often quite a hassle. Especially with GPU drivers, CUDA, cuDNN, etc. -- downloading, installing, recompiling is a time-suck at best. At worst, I've made mistakes with something not properly installed (but not realizing it) and the software works, but doesn't run as fast as it should. Then I'm not getting the full speed out of the hardware. This AMI takes care of all that guess-work and hassle. Super convenient!


Higher performance is great for our ML/DL applications

  • By Vilynx Inc
  • on 11/21/2016

Once we got the instance up and running, it all worked great. We rely on AWS for all our Machine Learning and Deep Learning needs so this higher performance AMI really helps us. Getting support for the multiple frameworks we leverage was key and the AWS support team has been great.


No nvidia driver or CUDA?!

  • By KK
  • on 10/06/2016

What's the point if no nvidia driver or CUDA? We need GPU to run those deep learning frameworks! We still have to recompile those frameworks if we install driver and CUDA.


it works as well as it can

  • By Har
  • on 10/05/2016

NVidia make their cudnn stuff proprietary, so that you have to register an account with them to download it. It's illegal to distribute it. Which is why this machine does not have the required libraries to be up and running.

It does provide instructions for doing it yourself, which is as PITA, but that is not Amazon's fault whatsoever.

After an hour I had mine up and running doing what I needed nicely. Thanks.