OMG! Now We Have Small Cheap Supercomputer Boards

A shaky start with a Kickstarter project has brought a small cheap supercomputer node to market. Stack them up and you have a lot of computing power in one box…
“Olofsson grabbed two 24-port Gigabit Ethernet switches and 42 of the Parallella boards to create a 42-node cluster that is about the size of a tower PC. It will cost around $5,000 and burn less than 500 watts (all in, including the three kinds of processing, memory, flash storage, and Ethernet ports).

Such a machine delivers around 1.1 teraflops of oomph, and by shifting to the 64-core Epiphany-IV would push that up to 4.3 teraflops. That’s not a lot of teraflops, and a bunch of GPU coprocessors can match that in a much smaller form factor to be sure. But the Epiphany RISC coprocessor is more than twice as energy efficient, according to Adapteva.”

see Adapteva ships Kickstarted baby supercomputer boards.

Adapteva is pre-ordering boards, kits, and connector-packages for October delivery. There is an SDK. It runs GNU/Linux, of course.

With the next version of Epiphany ARMed chip, and ramped up production, this will no doubt be a future entry in the Top 500 list of supercomputers. It’s price/performance is competitive as is. Mass production will make it very attractive. Just like that we have competition in the world of small cheap supercomputers.

This is yet another example of how FLOSS can make a small guy competitive in the world of global monopoly in IT. The big guys just cannot compete against FLOSS. Everyone benefits.

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology and tagged , , , . Bookmark the permalink.

3 Responses to OMG! Now We Have Small Cheap Supercomputer Boards

  1. Sean O'Connor says:

    You could simply buy 100 dual core ARM boards with 1GB RAM each and have an easy to program cluster to deal with real world problems. I feel that GPU solutions are too narrow in applicability and difficult to program. Unless that is, you have some simple key algorithm (suited to GPU’s) that you wish to apply again and again. Maybe the FFT or WHT in an artificial intelligence system. Then you could be onto a winner with GPU’s.

  2. dougman says:

    I was impressed with this, they want to place one million cpu cores in 30kW server rack.

    They should be placed at http://www.opencompute.org/ under micro-systems.

    Opensource hardware is the next revolution in IT.

  3. matchrocket says:

    A high school could build one of those.

Leave a Reply