If a defect can trap/store charge, is it a defect or a feature?

What I read (probably in Optics and Laser Europe) about holographic memories was quite interesting.

You know, if you stay at technologies similar to Flash (I mean, patterning small cells on a surface), it becomes more and more (extremely) expensive to get a small improvement. And by doubling capacities every 18 months, the one-atom size limit would be attained in 20 years only. Even the Millipede is limited!

Look, my first Flash card had 16MB, maybe 7 years ago. Now you find >16GB at the grocer's, this is >1.000 times more. But shrinking from 45nm to 0.045nm only gets you 1.000.000 times more cells, and we won't go that far.

So I wouldn't be surprised that thechnologies not evolved from semiconductors replace Flash. What about a fast course in optics and biochemistry...?

Don't kidding atoms size is 0.2nm, so 0.045nm is impossible. If sciencists don't find Ideas how to fit in bit-memory-cells into 3D structure, then in 2050 memory memory size growing stop at all (maybe grows +1MB/year and decrease farther). And from about 2030 all computers and memory evolution will exponentionaly slowdown and this proccess exist already now, but in future it will be more visible. You never in your live will see 100 TB hard disk or processor with 100 bilions transistors.

DavidD, your figures don't fit.

Shrinking down semiconductor technology will slow down much sooner having attained densities much bigger than you expect.

As for hard disks, we are on the best way to 100TB.

You know what is ironical, that 10TB flashHDD will have more than 100 trilion transistors, becouse 10TB=80Tb and for one bit need >1 tranzistors. Now for example exist 1 GB RAM, and each bit consist of 4-6 transistors, so 1GB=8Gb, 8*6=48 (or 32) bilions transistors is in 1 GB RAM. In intel new 4 cores processor is 820 milions transistors. 2 GB flash consist of at least 16 bilions transistors. Why in flash is much more transistors than in processor I particurly can explain, becouse processor working at high frenquency and need very much energy (100W) or it will burn if you insert more layers and then speed decrease, becouse of limitation speed of electricity. So if you will made processor with say 16 layers one on another then frenquency will decrease 16 times, becouse of electricity speed limits. So with 16 layers Intel procesor will be able work only on about 150 MHz (flash memory have many layers, about 20). RAM speed, frenquency is 400-800 MHz, and bus speed is 200-400 MHz (strange why not the same (twice smaller) as RAM) and so 1GB RAM has 16 chips, then one chip has ~2 bilion transistors and since each chip working about 4 times slower than processor, it consuming ~4 times less energy than processor, but number of chips on 1GB RAM is 16 and so 16/4=4 times RAM must consume more energy than 1 processor. But maybe there play role such phenomenon that at bigger frenquency energy consumption is less effiecent and RAM transistors are simpler conected with wires and so shorter distance between transistors... So flash memory based HDD can be possible, becouse it will work on very low frenquency and will consist of thousunds or milions layers, but now SDD (flash based hard disk) taking prety much energy with ~100 GB and so I have reason to think, that about 100 times SDD disk can be bigger becouse 50^2 (nm^2) / 5^2 (nm^2)=100 times more transistors can fit in the same size, so smaler than 5 nm technology very unposible and then SDD disks 100GB * 100= 10TB will be maded by doing smaller transistors, but farther by adding aditional 10 layers SDD disks will take 10 times more power. So maybe 100 TB is possible but it is close to limit, becouse aditional layers will made enormous energy consumption or will made very slow Those disks if they forking on enough big frenquency of about 100-1000 MHz... But those HDD disks which are with rotating disks, they will not have 100 TB, becouse they diameter is 0.1 m and they can have max 5 disks inside and 10 heads for each side. So S=10*pi*r^2=10*0.05^2 *3.14=0.0785 m^2. One bit can't take more than 10nm*10nm=10^{-8} *10^{-8}=10^{-16} m^2 space. So 0.0785/10^{-16}=7.85*10^14 bits ~800 Tb~100TB. So why usual HDD can't have more than 100 TB.

DavidD, the basic explanation is that you missed it.

A DRAM contains only ONE transistor per bit cell, not four or six.

Such regular cells can also be far better designed than a processor, don't need NMOS and PMOS near toanother, and use specialized manufacturing processes, so one DRAM transistor uses far less space than a processor transistor.

Put this together, and you get the whole explanation.

By the way, you should have a look at what is meant with bus frequency, DDR, DDR2 and quad-pumped. It might clarify some discrepancies.