HP boffin: Honey! I shrank the PC. To nanometre size, dammit

".....could distinguish it from Dell, Cisco and IBM....."

Don't be silly, hp is not likely to go back into the processor game. It's more likely to license the tech to other CPU manufacturers, such as best buddy Intel for Xeon and Itanium, which means Dell and CISCO and IBM will have equivalent offerings almost as soon as hp do. Maybe hp will have some advantage from knowing more about how the tech works, allowing them to design better systems around the new chips or get products out the factory door first, but I suspect the license revenue model is where hp will look to make money off this. And it's more than likely it will be licensed into Power seeing as IBM don't have a snowball's chance in Hell of getting any competing tech ready in time to compete with Xeons and Itaniums packing memristor tech. Phase Change Memory aka PRAM is aimed at flash replacement and doesn't like hot environments like the inside of p-series ovens.

exciting

but the proof will be in memory access latency and OS support. Both file IO and virtual memory manager, and probably other bits of kernel, are going to be affected by removal of a barrier between RAM and nonvolatile storage, in connection with absolute minimum of resources available.

Naturally, It won't be an x86 architecture CPU on board....Why make this easy?!

So, we could have a great idea, with say Itanium3 attached, wrecking its prospects.

I think this could be cool, if it works and the fab can cook more than 1 of them per wafer batch.

Mind, there is a rumor Intel and Micron are going down a similar path.

Smart way to go, especially in the long run

The nanostore design appears to place energy efficiency above data retrieval rates

Also massively distributed processing.

I can think of something else that works this way, that's been honed by hundreds of million years of evolution. I'm sure you can too ;-)

"smart memory" rides again. The late 80's are back.

Back then various LSI conference proceedings were full of architecture to link simple (usually bit serial) processors to individual storage chunks down to the bit level, promising massive processor/memory bandwidth. Image processing was a popular theme but rock strata reconstruction (IE looking for oil) was in there too.

IIRC only the ICL design made it into production. I think N. Carolina @ Chapel Hill were keen on it as well.

So the density has improved, the chunks got bigger and the processors will presumably be full 32 bits.

Didn't happen then. Not convinced it'll happen now.

But yes, it could play crysis.