A Supercomputer On A Chip For Everyone

Every now and then, an exciting Kickstarter project comes along that I want to support. Some succeed, some fail. The newest entry is developing an open source supercomputer-on-a-chip that is smaller than a credit card and suitable for parallel programming applications. For a $99 USD pledge, you can get a developer board and the open source software to play around with when it comes out.

The closest I ever got to parallel programming was running multiple threads concurrently in Java (i.e., one thread for main loop, one thread for user interface, and different threads for processing data) on a single-core CPU while taking programming courses at San Jose City College. Parallel programming is breaking the data into identical pieces that can be processed on multiple  CPUs or computers in a network from a master program.

For years I wanted to build a micro beowulf cluster that consisted of five micro-ATX boards in a stacked 16″ cube. Although the price for hardware has gotten lower over the years, it’s still expensive to get identical hardware to build it out at the same time. Even as a 16″ cube, a micro beowulf cluster still takes up space and produce heat. A supercomputer-on-a-chip doesn’t have to take up more space than a small form factor (SFF) computer. At $99 USD each, that’s a steal.