Is there any diminishing return on performance vs. power when we continuously add more core to the processor? Frankly, I don't know if a quad-core is really twice better work than a dual-core so I just this may be a bit crazy to consume even more and more power to get a little gain on (computation) performance. I wonder if this is just a gimmick than a real performance improvement. However, with the poor battery life of the new smart phone, I prefer to have some more breakthrough on burning less power!
The theory actually works the other way. Power scales as square of freq, so if you double the cores you can in theory half the frequency and end up with half the (total) power.
Now, all you need to do is to figure out how to write the software. But with each core just a few sq mm of Si, why not just make them and see if anyone can solve how to use them? And when nothing is running that needs them, just shut down the idle cores.
"Though I don't know what to do with 8 cores!"
its seems simple enough :)
given this is eetimes, get your soldering iron out and make a new 8 core http://trimslice.com/web/ type device PCB for 2013 and sell it as a real and viable low power desktop unit as http://blogs.arm.com/smart-connected-devices/858-galaxy-note-ii-great-innovation-and-the-marmite-of-smartphones/ makes reference to "this Gigaom article and video on using the Galaxy Note II as a desktop computer" there as one use for this/other 8core cortex SOC in 2013
arm cortex will get something like 80% performance per core added just like x86.
but OC they also have the CoreLink Cache Coherent Interconnect to help them out too so more spare cycles use use http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0470b/index.html