Adapteva’s Parallela board -- small but powerful
The last SBC in this list comes from Adapteva, and isn’t even available yet. However, the specs on it are promising. It is even being promoted for use in parallel computing. The company successfully crowd-funded the board back in 2012 and has shipped the SBC to its backers. Due to the high demand for the board, it’s currently not available until the company can make more.
Adapteva’s Parallela board may be small, but it packs a powerful punch.
So why is the board in such high demand? Simple, it’s great for parallel computing and can easily be strung together to create a monster of a supercomputer without spending a small fortune to do so. The board features some of the same amenities of the others, including one GB of RAM, two USB 2.0 ports, Ethernet, and an HDMI connection, but that’s where the similarities end. The board is outfitted with Zynq-7010 dual-core ARM A9 CPU and either 16- or 64-core Epiphany-IV RSIC coprocessors to handle the heavy lifting. Simply put, the board is a supercomputer unto itself.
Adapteva says that its 64-core version is capable of delivering 90 GFLOPS of performance and is comparable to a theoretical CPU running at 45 GHz or 64 cores running at 700 MHz each. The board (if you can get it) uses Ubuntu for its OS and comes with a software stack that includes Epiphany development tools such as a C compiler, a multi-core debugger, and OpenCL (among a host of others) for those needing a development platform.
At $99 per board (for the base platform), the board is more on the expensive side, however it’s relatively compact (about the size of a motherboard with 64 cores) and incredibly powerful. While these boards don’t necessarily compare with supercomputers that cost millions of dollars, they are an incredible alternative for home users who need to compile data and don’t want to sell their internal organs to get the ability. Developers, makers, and hobbyists alike are continually turning to SBCs for their projects. The trend will continue to grow, especially when they can purchase off-the-shelf hardware and turn it all into a data-crunching beast of a machine.