Here’s a surprising name on the new list of the world’s fastest supercomputers: Amazon Web Services.
Looking to promote its latest offering in High Performance Computing — dubbed the Cluster Compute Eight Extra Large, or CC2 for short — the Seattle company’s cloud computing team has submitted its benchmark speed results to the keepers of the closely watched Top500 list — becoming the proud owner of the world’s 42nd fastest supercomputing system.
Amazon’s speed of 240.09 teraflops (more than 240 trillion floating point operations per second) pales in comparison to the 10510 teraflops (a.k.a. more than 10 petaflops) registered by the reigning champ, the K Computer at Japan’s RIKEN Advanced Institute for Computational Science.
But the Seattle company isn’t so much trying to compete with the top supercomputers as much as it’s trying to bring supercomputing to the masses — offering a cloud-based solution to scientists, engineers and researchers who need to process huge sets of data.
Writes Amazon’s Jeff Barr in a post this morning: “On a somewhat smaller scale, you can launch your own array of 290 CC2 instances and create a Top500 supercomputer (63.7 teraFLOPS) at a cost of less than $1000 per hour (perhaps a lot less, depending on conditions in the Spot Market).”
The latest version of the Top500 list was released in conjunction with the big SC11 supercomputing conference, taking place this week in Seattle. Seattle’s Cray supercomputer company is all over the list, as well, starting with the No. 3 fastest supercomputer (1759 teraflops), a Cray installation at the Oak Ridge National Laboratory.
In the future, however, Amazon will need to step it up to keep its ranking. The researchers behind the Top500 list predict it will consist entirely of machines running at a petaflop or more by 2016.