Last week we reported that HPC company Cycle Computing built a 10,000-core cluster on the Amazon EC2 cloud service. Cycle CEO Jason Stowe boasted that the cluster was big enough to make the list of the world's Top 500 supercomputers -- if only it had been subjected to the required speed test. Well, it turns out there already is a cloud-based supercomputer on the Top 500 list -- and it was built by Amazon itself.
CLOUD CLUSTER: 10,000-core Linux supercomputer built in Amazon cloud
Ranking as the 231st fastest supercomputer in the world, Amazon's cluster contained 7,040 processing cores and achieved speeds of 41.82 teraflops. The cluster ran Linux, used Intel Xeon X5570 processors and a 10 Gigabit Ethernet interconnect. The supercomputer was de-provisioned soon after running the test.
Amazon's cluster was on the November list of the Top 500 supercomputer sites, but we didn't notice it at the time.
One of its builders, James Hamilton of the Amazon Web Services team, declined an interview request from Network World but explained the project in some blog posts last year.
"The Top 500 Super Computer Sites list was just updated and AWS Compute Cluster is now officially in the top ½ of the list," Hamilton wrote. "That means when you line up all the fastest, multi-million dollar, government lab sponsored super computers from #1 through to #500, the AWS Compute Cluster Instance is at #231. One of the fastest supercomputers in the world for $1.60/node hour. Cloud computing changes the economics in a pretty fundamental way."
CLOUD INNOVATION: 10 cloud computing companies to watch
Amazon officials decided to run the Top 500 test -- known as the Linpack benchmark -- to test out new cluster compute instances for HPC applications they were offering to customers. Amazon engineer Matt Klein ran the Linpack on an 880-server cluster.
"It's a good test in that it stresses the network and yields a comparative performance metric," Hamilton wrote. "I'm not sure what Matt expected when he started the run but the result he got just about knocked me off my chair when he sent it to me last Sunday. Matt's experiment yielded a booming 41.82 TFlop Top-500 run."
Hamilton continued to note that "This is phenomenal performance for a pay-as-you-go EC2 instance. ... It also appears to scale well which is to say bigger numbers look feasible if more nodes were allocated to LINPACK testing. As fun as that would be, it is time to turn all these servers over to customers so we won't get another run but it was fun."
Follow Jon Brodkin on Twitter: www.twitter.com/jbrodkin
Read more about data center in Network World's Data Center section.