Chip-company Adapteva has built its first parallel-processing board for Linux supercomputing. Those who had backed the Parallella Kickstarter campaign will be getting their hands on the processor board, named the Parallella, by summer. The company has given us the first proper look at the complete Parallella processing board. Judging by the pictures, Adapteva does indeed seem to live up to its promise of having a credit card-sized supercomputer.
The Parallella has been built with the purpose of making supercomputers more powerful. While Linux was the operating system of choice for supercomputers, building a fast one has always been a tricky issue. According to the company, "The goal of the Parallella project is to democratize access to parallel computing."
It is essentially a credit card-sized parallel-processing board, much like the Raspberry Pi. It has a dual-coore ARM A9 processor and a 64-core Epiphany Accelerator Chip. Along with this, it has 1GB of RAM, a microSD card, two USB 2.0 ports, 10/100/1000 Ethernet and an HDMI port. In theory, the processor board should be able to give a performance of about 90 GFLOPS, which is the same performance as that given by a 45GHz CPU. All the while, the board will consume only about 5 watts worth of power under typical work loads.
The first Parallella boards have been built
Th Epiphany multicore chips consist of a scalable array of RISC processors that are programmable in C or C++. They are connected together with a fast on-chip network within a single shared-memory architecture.
The operating system of choice for the board will be Ubuntu 12.04
The company states that the Parallella follows three principles: Open access, open source and affordability. The product has no NDAs or special access needed. The architecture and SDK documentation is available online. Taking the affordability principle, the Parallella is set to cost less than $100 (approximately Rs 5,403).
The Parallella can potentially be used for many different things. For consumers, it can be used as an energy-efficient computer media box. The imaging capabilities include face detection/recognition, finger print matching, object tracking, stereo vision and gesture recognition, among other things.
It also has some interesting alternative uses, such as the ability to power a robot, space electronics and multi-sensor inertial navigation.
The main use will undoubtedly be high performance computing. With it, you can have real-time Internet stream analytics, real-time market analytics, soft encryption engine, code breaking and data logging, among other uses.