Amazon Web Services wants to move further up-market and service more of the HPC crowd so it's publicly beta testing a Cluster Compute Eight Extra Large (CC2), its biggest, most powerful cloud service to date.

The thing has 16 Xeon cores (a pair of eights with Hyper-Threading enabled), 60.5GB of RAM, 3.37TB of storage and 10 gigabit Ethernet for cluster communication. Users can have it with Linux or Windows Server 2008 R2.

It costs $2.40 an hour per instance or $4,146 for a reserved instance on a one-year contract. Cheaper pricing might be had on the EC2 Spot Market.

An array of 290 CC2 instances good for 63.07 teraFLOPS would run $1,000 an hour. Amazon fashioned a Linux cluster of 17,024 cores (1,064 instances) that ranked 42 on the latest Top 500 list just out. It hit 240.09 teraFLOPS.

AWS figures the widgetry is good for physics simulations, seismic analysis, drug design, genome analysis, aircraft design, financial services and some business computing and analytics applications using its Elastic Map Reduce and Hadoop widgetry.

CC2 is currently only available in Amazon's data center in Virginia.

Harvard Medical School has been using Amazon's existing Cluster Compute Instances, introduced last year, for genomics. Amazon has lowered the price of the existing CC1 instance to $1.30 an hour.

See http://aws.typepad.com/aws/2011/11/next-generation-cluster-computing-on-amazon-ec2-the-cc2-instance-type.html.