The november issue of the Top500 Supercomputer sites has just been released and there is a new entry in at number #231.
|231||Amazon Web Services
|Amazon EC2 Cluster Compute Instances – Amazon EC2 Cluster, Xeon X5570 2.95 Ghz, 10G Ethernet / 2010
The AWS EC2 Cluster Compute Instance is now officialy a supercomputer! AWS’s cluster compute instances are based on dual quad-core Nehalem chips and have 10Gbit ethernet interconnect.
In the table above they quote 7040 cores which means that they used 880 cluster compute nodes for the test.
This has means that any company requiring supercomputing processing power can afford to rent the entire cluster of 8808 nodes from Amazon for less than $1500. All they require is a credit card.
The implications for this are huge. It opens up supercomputing to the masses.
Amazon launch GPU processing
Today has been quite a day of news from the Amazon camp. They have also announced the immediate availability of Cluster GPU instances for HPC processing. A lot of our customers have expressed an interest in trying out their algorithms on GPUs and now they can do so without having to make a huge investment. The new instance types are based on the nVidia Fermi architecture as well as “normal” dial quad-core Nehalem CPUs.
|Cluster GPU Quadruple Extra Large Instance|
|22 GB of memory|
|33.5 EC2 Compute Units (2 x Intel Xeon X5570, quad-core “Nehalem” architecture)
2 x NVIDIA Tesla “Fermi” M2050 GPUs
|1690 GB of instance storage|
|I/O Performance: Very High (10 Gigabit Ethernet)|
|API name: cg1.4xlarge|
They are currently ony available in us-east-1, the North Virginia region.
In the near future we will be writing an entry on our technical blog going into the technical details.