Supercomputers have been a symbol of achievement for corporations, government entities, and academic institutions for decades. They have been used for everything from predicting market trends to cracking codes. Over time, these power-hungry machines have grown in scope, featuring thousands of CPUs and workstation GPUs to crunch massive amounts of data. They require massive amounts of electricity and cooling to keep them going.
Unfortunately for us, these machines are so excessively big and costly that they remain fodder for science fiction -- and forget running one out of your home.
The IBM Blue Gene/P supercomputer installation at the Argonne Leadership Angela Yang Computing Facility, located in the Argonne National Laboratory in Lemont, Ill., December 2007.
(Source: Argonne National Laboratory)
However, since the SBC (single-board computer) boon of late, people see the possibility of building a pretty powerful supercomputer that won’t break the bank.
Nvidia's recently released Jetson TK1 Development.
$192 Nvidia’s Jetson TK1 Development Board
Nvidia’s Jetson TK1 Development Board is an attractive option for home-based supercomputing. There are dozens of SBCs on the market, however three really stand out among the crowd and make great data compilers. First up is Nvidia’s recently released Jetson TK1 Development Board, which features the company’s long-awaited Tegra K1 SoC. The board is actually a complete computer all on its own and features the same Kepler core found in the company's current-gen gaming cards and professional graphics workstations.
To put that into perspective, the SoC is packed with a mind-blowing 192 cores. It was shown running the GPU demanding Unreal Engine 4 at CES back in January. The board also packs 2 GB of RAM, 16 GB of storage, a mini PCI-E slot, an HDMI port, and an SD/MMC connecter. If that wasn’t enough, it also features USB 2.0/3.0 ports, a RS232 port, a SATA port, and onboard ALC5639 Realtek audio with Mic and Line connections. Out of the box, the board comes preloaded with Linux for Tegra, OpenGL 4.4, CUDA, and the VisionWorks tool kit for those who want to use the board for developing applications and devices. On the other hand, multiple boards can be strung together to create a powerful home-based supercomputer for just about any need. At roughly $192 a pop, that won’t break the bank!