With all due respect to these continuing amazing achievements, I'm no longer blown away by the ever-decreasing combination of price, size, power consumption against ever-increasing GPU, I/O & arithmetic performance. It seems clear this is just going to keep on at the same amazing rate, for quite a while.

What I wait eagerly to be blown away by is innovative uses[1] for these supercomputers that fit on your thumbnail and could be powered by the kinetic energy you generate when you walk (excuse the hyperbole.) Bring on the applications!

By which I mean applications of technology, not touchscreen apps. Low-power server farms & netbook-/iPad-alikes are fine and all, but I'm excited about what else gets created when excess computing power is essentially throwaway cheap and super portable...

"Ubiquitous computing" has been coming for a while now, I can't wait to actually see it!

</rant> :)

[1] iPads and iPad-derived ideas aside.

You need innovation at all layers of the stack. imo our engineer friends who work on lower layers of the stack dont get all the press and love they deserve.

If my math is right, 5.5 W means you could power this chip from a pair of high-end rechargable AA:s for about an hour. That's pretty cool, for something with this level of performance.

I assumed 2400 mAh cells at 1.2 V; two in series gives 2.4 V, powering a 5.5 W load at 2.4 V requires a current of 2.3 A which said cells should be able to deliver for one hour (2300 mAh).

Of course, this skips any contemplation of what voltage the chip actually wants, which is probably lower than 2.4 V, and thus means some conversion (with losses, etc) needs to happen.

I know everyone's fed up with the bitcoin articles, but I still wonder what this means for mining. Is there a CE or architecture guru who might be able to speculate how this architecture would perform? The reduced power footprint could make personal mining more viable in the long term as FGPU clusters start to take over.