Deep Learning AMI
Developer Guide

Release Note Details for Deep Learning Base AMI (Amazon Linux) Version 1.0

AWS Deep Learning AMI

The Deep Learning Base AMI are prebuilt with CUDA 8 and 9 and ready for your custom deep learning setup. The Deep Learning Base AMI uses the Anaconda Platform with both Python2 and Python3.

Highlights of the Release

  1. Used Amazon Linux 2017.09 (ami-8c1be5f6) as the base AMI

  2. CUDA 9

  3. CuDNN 7

  4. NCCL 2.0.5

  5. CuBLAS 8 and 9

  6. glibc 2.18

  7. OpenCV 3.2.0

Python 2.7 and Python 3.5 Support

Python 2.7 and Python 3.6 are supported in the AMI.

CPU Instance Type Support

The AMI supports CPU Instance Types.

GPU Drivers Installed

  • Nvidia 384.81

  • CUDA 9.0

  • CuDNN 7

Launching Deep Learning Instance

Choose the flavor of the AMI from the list below in the region of your choice and follow the steps at:

AWS Deep Learning AMI Developer Guide

Deep Learning AMI (Amazon Linux)

This AMI is available in the following regions:

  • US East (Ohio): us-east-2

  • US East (N. Virginia): us-east-1

  • US West (N. California): us-west-1

  • US West (Oregon): us-west-2

  • Asia Pacific (Seoul): ap-northeast-2

  • Asia Pacific (Singapore): ap-southeast-1

  • Asia Pacific (Tokyo): ap-northeast-1

  • EU (Ireland): eu-west-1

Test Environments

  • Built on p2.16xlarge.

  • Also tested on p2.xlarge, c4.4xlarge.

Deep Learning Base AMI (Amazon Linux) Known Issues

  • Issue: Versions of pip and Python are not compatible for the Amazon Linux Deep Learning AMI (DLAMI), specifically pip, which is expected to install Python 2 bindings, but instead installs Python 3 bindings.

    This is a known problem documented on the pip website. A future release will address this issue.

    Workaround: Use the relevant command below to install the package for the appropriate Python version:

    python2.7 -m pip install some-python-package

    python3.4 -m pip install some-python-package

  • Issue: The MOTD has an incorrect link to these release notes.

    Workaround: If you made here, then you already know.

    A future release will address this issue.