6.5 TFLOPS of Cryptographic Power in Your Palm?

Tuesday, October 09, 2012

Damion Waltermeyer


Coming soon to your hand, Massive Parallelism? That depends on you.

We’ve all had that pesky problem of not having a powerful number cruncher handy when we need it.

That may soon be solved by Adapteva Inc.   They are attempting to consolidate their accelerator boards into 28nm chips; and because of the cost have turned to kickstarter.


What does this mean for us?  Massive Parallelism for cracking, hashing, log correlation and other large data applications.  

The power of parallelism cannot be overstated when it comes to these sorts of applications. Rather than go for a single powerful CPU, Adapteva has chosen to use a Dual ARM CPU with low power accelerator cores. This allows them to scale up cores cheaply and rapidly with little increase in watts or heat.

I wrote out this long example of Parallelism, but then I remembered a much more succinct explanation.


The Parallella is about the size of a credit card or Arduino or Raspberry Pi setup, but a whole lot more powerful. It has a full C and C++ library available for your exploits and experiments.

The initial 16 core 28nm version gives you 25GFLOPS at 5 watts for the whole board.  Supporters who provide at least $99 will receive one of these.

The 64 core 28nm version that is the stretch goal gets you 90 GFLOPS for about 5 watts at 8.2mm(2), supporters who provide at least $199 will receive one of these if they meet the stretch goal, if not they receive two 16 core setups.

Their next few steps after that go up rapidly. They expect to have a 4096 core board in the next few years that gets you 6.5 TFLOPs, just under a hundred watts and the processing portion would fit in about a square inch.  So while the powersupply would make it a bit larger, in the next few years you might be able to fit 6.5Tflops inside an altoid tin(assuming you can find a way to dissipate the heat from 100W).   

None of this will happen without your support.  Consider what you spend for a Tablet, a phone or even a graphing calculator and then see if you can’t find a little bit of cash for an open source Parallel processing platform.  

Disclaimer: Nothing. No Connection with these guys other than thinking this idea is awesome.

Post Rating I Like this!
The views expressed in this post are the opinions of the Infosec Island member that posted this content. Infosec Island is not responsible for the content or messaging of this post.

Unauthorized reproduction of this article (in part or in whole) is prohibited without the express written permission of Infosec Island and the Infosec Island member that posted this content--this includes using our RSS feed for any purpose other than personal use.