Launch an Amazon Deep Learning Amazon Machine Images (AMI) instance
- Log in your AWS account. This will go to the AWS Management Console.
- In the AWS Management Console, click on the Services menu, then click EC2.
- On the upper right corner, click the region button, select US East (N. Virginia).
- Click Launch Instance.
- On the Step 1: Choose an Amazon Machine Image (AMI) page, scroll down to Deep Learning AMI (Amazon Linux).
Confirm that it is the Amazon Linux version, not the Ubuntu or Windows version. There are multiple versions, so select the highest version.
- For Deep Learning AMI (Amazon Linux), click Select.
The Deep Learning AMI is designed to provide a stable, secure, and high-performance execution environment for deep learning applications running on Amazon EC2. It includes popular deep learning frameworks, including MXNet, TensorFlow, PyTorch, Keras, Chainer, Caffe, Theano, and CNTK.
- On Step 2: Choose an Instance type page, for Filter by, select GPU instances.
- Select p2.xlarge.
- Click Next: Configure Instance Details.
- On the Step 3: Configure Instance Details page, keep the default settings.
- Click Next: Add Storage.
- On the Step 4: Add Storage, keep the default storage settings.
- Click Next: Add Tags.
- On the Step 5: Add Tags page, click Add Tag, then:
- For Key, type: Name
- For Value, type: Deep Learning Development
- Click Review and Launch.
- On the Step 7: Review Instance Launch page, review your instance settings and click Launch.
- On Select an existing key pair or create a new key pair window, if applicable, select Create a new key pair from the dropdown menu. Assign a key pair name (for example, DeepL) and download the private key file in a secure and accessible location. Amazon EC2 gives you a PEM file. So, you would need to convert it to PPK files later for Windows SSH connectivity to the instance.
- Click Launch Instances.
- The launch could fail if you choose regions other than US East (N. Virginia) at the first beginning. You could either change the region to US East (N. Virginia) and retry or launch it after you get an email saying that your account is now fully verified in that region. This verification will take several minutes.