SCIENCE

Cycle Computing has announced that as a part of its continuing partnership with Amazon Web Services, it now provides users with access to Amazon's new Cluster GPU Instances, designed to deliver the power of GPU processing in the cloud.

Cycle Computing has had early access to these new GPU node types and benchmarked them against customer's existing in-house GPU hardware. Jason Stowe, founder and CEO of Cycle Computing, is available this week at SC10 in New Orleans to discuss the new GPU instances and the benefits of accessing them via CycleCloud, including:

• How to provision a cluster: Utilizing CycleCloud's ability to provide HPC and HTC clusters in the cloud, it is possible to create Condor/Torque/SGE clusters on Amazon EC2 using the new GPU instances in minutes rather than days. A demonstration of how to provision a cluster is also available: CycleCloud GPU-enabled HPC clusters in the cloud using Amazon EC2.
  
• Benchmarking results: Cycle Computing approached a Fortune 500 life science and a finance/insurance company, both of which develop and use their own GPU-accelerated software, to run their applications on the GPU-enabled Cluster Compute nodes. Stowe can discuss the results of specific GPU tests and the comparison to client's in-house hardware.
  
• Internal hardware vs. virtualized, cloud-based hardware: With these new nodes, the line between internal and cloud-based hardware for HPC using GPUs has been blurred.