If you've been following the story of high performance computing on AWS, you'll see that we added instances with general purpose GPUs just under two years ago. In that time customers have been using these powerful, multi-core processors for a broad range of applications which can take advantage of the massive parallel computing capabilities, from rendering to transcoding, computer vision to molecular modeling.
Today, we're making that same functionality available more broadly:
- The Cluster GPU instances (cg1.4xlarge) are now available in the EU West (Ireland) region. That means that customers who have data or additional supporting resources deployed in this region can now accelerate their projects with these high performance instances.
- Additionally, today Cluster GPU instances are also available to launch inside Amazon VPC. If you're using VPC to build public and private subnets for your computational resources, you can add GPUs into the mix in just a few API calls.
These instances are available on-demand, as reserved capacity or via the spot market for even more bang for your buck.
You can fire up a GPU instance and run some molecular dynamics computations across all those cores in just a few clicks:
This CloudFormation template will launch a GPU instance with a custom made AMI from the OpenMM team at Stanford University, who recently ran a molecular modeling workshop using this instance type to help get their students access to all those cores quickly and easily. You can find instructions, test code and other examples in the home directory of you new instance.
Let us know how you're filling these up.