SOURCE: Cycle Computing

Cycle Computing

November 11, 2014 13:32 ET

Cycle Computing Software Deploys 50,000 Cores in 23 Minutes -- Largest Fortune 500 Cloud Cluster Computing Run

729 TeraFLOPS Cloud Cluster Reduced 30-Day Computer Simulation to 8 Hours

LAS VEGAS, NV--(Marketwired - Nov 11, 2014) - Cycle Computing, the leader in cluster Cloud computing, announced today it has created the world's largest enterprise Cloud computing run by leveraging more than 70,000 cores on AWS for a Fortune 500 manufacturing company. The manufacturer tapped Cycle Computing's software to reduce the amount of time required to run a workload that took 30 days down to just 8 hours. The workload was accelerated by a Cloud cluster with 70,908 cores that measured 729 teraFLOPS (trillions of floating point calculations per second at peak), and would be comparable to the 63rd fastest supercomputer in operation today, all for just $5,593.94. The impact to the business is to reduce time to market of new and innovative products.

Enterprise-speed Cloud Clusters
Dubbed the Gojira run, Cycle Computing was able to quickly and securely create the Cloud cluster tapping into 70,908 cores over three AWS regions. It took Cycle Computing just three days from the initial request to run a workload this large, until the workload was completed, and the manufacturing company's engineering team had its results.

"Today we have demonstrated not only the scale of what's possible, but also the agility customers have to tackle problems in days with resources that are 100 times greater than what they have in-house," said Jason Stowe, Cycle Computing CEO. "Since 2010, Cycle has encouraged using Cloud to ask the right questions, regardless of scale. Today, this Fortune 500 use case shows how enterprises are adopting Cloud and engineers who rely on computer simulations are truly no longer limited by computing infrastructure."

Record Cloud Cluster Computing Powered by CycleCloud
CycleCloud is the leading software for creating Technical Computing and HPC (high performance computing) clusters in the Cloud, from modest sized systems of 64-6,400 cores to systems that rank as some of the fastest computers in the world (156,000+ cores). CycleCloud makes it easy to deploy, secure, automate, and manage running calculations dynamically -- in most cases with a single click. CycleCloud takes the delays, configuration, administration, and sunken hardware costs out of HPC clusters. Now, engineers, researchers, and quants can use CycleCloud to size the infrastructure to the technical question or computation at hand.

Gojira by the Numbers

Metric   Count
Compute Hours of Work   619,748 hours
Compute Years of Work   70.75 years
Design Count   ~1 Million drive head designs
Run Time   8 hours, not 30-days in-house
Applications Used   MRM/MatLab, CycleCloud, Chef
Max Scale (cores)   70,908 AWS cores, 3 regions
Max Scale (instances)   5,689 Spot Instances at peak
Computing Power   729 TeraFLOPS rPeak, more than #63 on Top500 rPeak
Infrastructure Costs   AWS Spot Instances: $5,594

About Cycle Computing
Cycle Computing™ is the leader in Cloud cluster computing, which enables greater access to computing over the Cloud. Cycle Computing software CycleServer™, CycleCloud™, and DataMan™, represent the enabling force that helps users easily connect to greater computing power using public and private clouds. The company believes that utility access to HPC will lead to a new era in accelerated invention and discovery. Since 2005, starting with three initial Fortune 100 clients, Cycle Computing has grown to deploy proven implementations at Fortune 500s, SMBs and government and academic institutions including JP Morgan Chase, Purdue University, Pfizer, The Aerospace Corporation, and Lockheed Martin.

Contact Information

    Brad Rutledge
    Cycle Computing
    Email Contact