The *really* big thing in memory, in production now by Intel in 128 
Mbit test devices aimed at niche applications, and liable to get going 
in the general market inside the next 5 years, maybe rather less, will 
(probably) be GeSbTe (or similar materials) based "phase change 
memory" which will "converge" several memory types (eg NOR Flash 
applications and DRAM), bring non volatility to traditional volatile 
memory areas,  reduce price per bit, overcome the pending sub 0.25 
micron silicon bottleneck, increase speeds by an order of magnitude, 
have RAD hard capabilities, and ... .  Doesn't yet look like it will 
make coffee or walk the dog. [[If it replaces NOR Flash at a 
significantly lower price and increased speed and ... it will also 
compete against and probably eliminate NAND flash]]

The obvious URL, but Gargoyle knows much more

        http://en.wikipedia.org/wiki/Phase-change_memory

With switching speeds around 2nS you can hope for a Flash memory 
replacement with over 3000X speed rating [!!!]. Sell your shares in 
other technologies now :-).

The technology was first investigated in the 1960s (first patent late 
1966) and even Gordon Moore wrote a paper on it in 1970 but it looks 
like its day has finally almost come. It had hurdles to overcome which 
were non compelling while silicon could be scaled with relative ease 
but the pending physical limits for silicon have spurred the required 
development effort. Everyone who is anyone in the memory world is 
working on it and no doubt a few startups as well. Massive paradigm 
shifts which technology changes like this can represent can be the 
opportunity for startups to gain a foothold in otherwise hard to 
penetrate markets and can see slow innovators die. It's also possible 
for such systems to fail to live up to their promises and burn 
everyone involved. (Bubble memory, Josephon junction stuff, CCD 
storage (except for the 1 MB or so chips used to acquire and store 
images for brief periods after you push the "shutter release" which 
seem to be in demand all over), ...).(Such technologies may retain 
lucrative niche application areas but fail to supply mainstream 
volumes).

In a physically moved hard disk replacement possibility it looks like 
densities of around 128 GB/square inch of memory surface are 
achievable - limited mainly by "platter" positioning abilities and 
sensor sizes.

Basic principle is phase changing of a suitable "glass" between 
amorphous and crystalline states with switching speeds of a few 
nanoseconds.




        Russell


-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist