Toward an exaflop computer

We all know that a lot of our technologies are driven by demand – without the computationally expensive experiments at the LHC, distributed computing itself wouldn’t exist in the form it does today.

Looking to the future, there is one scientific project on the horizon to watch, and that’s the Square Kilometer Array (SKA).

Set to be fully operational in about 12 years or so, the SKA is a next-generation radio telescope made of thousands of extremely sensitive dish antennas spanning about 5,000 km in the isloated regions of either Southern Africa or Western Australia (the site will be chosen in 2012).

Needles to say, the SKA will generate huge volumes of data: many exabytes per year, perhaps up to an exabyte of data every 13 days.

Some of this data will be easily mapped onto a distributed network of computers, for example small sections of the sky that independent processors can search for signs of pulsars. Other tasks will require a dedicated HPC cluster, such as simulating galactic evolution.

A recently released report looked at the computing requirements of this endeavor, and whether cloud computing could be used as part of the strategy, including desktop cloud computing – using spare CPU cycles in a similar fashion to the SETI@Home model.

You can read the full report here: Cloud Computing and the Square Kilometer Array