Slashdot Mirror
Researchers Achieve Amazing Memory Density
Mr. Fahrenheit writes in with a Wired story on research out of Arizona State, where researchers have "developed a low-cost, low-power computer memory that could put terabyte-sized thumb drives in consumers' pockets within a few years... The new memory technology — programmable metallization cell (PMC) — comes as current storage technologies are starting to reach their physical limits." PMC involves the on-demand creation of copper nano-wire bridges. It's said to promise memories that are 1/10 the cost and 1/1000 the power consumption of conventional Flash memory. Three memory manufacturers have licensed the technology and the first chips are expected on the market in 18 months.
FTA: "Kozicki says the process is like condensing a crystal from a solution, except that the process is almost infinitely reversible." Remember that gargoyles episode where like half of australia gets covered in nano crystals? That's what your room looks like after a drive failure.
The problem with how memory is that it gives developers no incentive to optimize code to run it faster/better/smaller other then small speed boosts. 1 TB of storage would be nice, but if it means that I have to download 300 GB for a program or a Linux distribution with the same speed of 1 MB/second it would take forever or say a 7 MB web site. We need to see an increase in Internet speeds at affordable prices first before we go overboard with physical storage.
Um... there ARE other uses for lots of storage, you know? Say, backing up in the field after spending a week shooting a couple thousand images per day with a digital camera that writes 50mb files?
Video?
Multi-track digital audio?
It isn't always about Linux distros, you know?
Don't disappoint your bird dog. Go to the range.
I went to the website http://www.axontc.com/ . and found following description;
"Key Benefits
PMCm has a number of unique attributes that make it a highly attractive component for future systems on silicon:
Operation at low voltages ( 0.3 V)
High speed write and erase operations
( 30 ns)
Low energy to change state ( 1 pJ)
Physical scalability to tens of nm
Easy integration with IC logic circuitry
Operation as a low refresh-rate DRAM or as a true non-volatile memory with high endurance (based on the programming mode).
These features define a class of devices that are essential for projected electronics systems and which will be difficult to realize using developed versions of today's circuits. "
Hope that answers some of your questions
Did you even read the article? It talked about the technology being 1000 times more energy efficient than what's currently in use. This isn't actually that hard to believe. The statement from the article that the process is nearly reversible speaks to the thermodynamically reversible case in which infinitesimal amounts of energy are used in each step of a reaction. Considering they're talking about assembling copper ions into nanowires, I think the speed will be quite reasonable. I mean, on the atomic level, reactions happen at the femtosecond scale and the actual solution is probably also going to be optimized for ion mobility. In terms of reliability, there isn't much you can do to screw something up at that scale if the thing is produced properly to begin with. I imagine if anything were to 'wear out', it'd be the array to which these ions affix to form the bridge.
I'm really interested to see how this technology matures, but as for now it sounds like a perfectly natural progression from what we have.
Looking around the web, I see a lot of stuff from around the year 2000 about how the paperless office is a myth and paper use increased for the previous 20 years, but more recently we seem to have turned a corner. Not that we'll be truly paperless, but the growth in demand for paper is less than GDP growth and usage per worker is actually falling. There's even a quote from a paper company in there.
I can not remember where the web page is located that had the info I used. Google return several pages, but only touch on "link-dest" for a short paragraph. Rsync docs are your best bet for farther info.
Actually the technology they present is quite realistic. I did look at the article and the technical paper available on their website. They are talking about building the conducting bridge in an heavily doped material where the conductible material are sphere of approximately 20nm with a spacing of approximately 2nm(approximately 2 atoms diameter).Submitting such a solution even if it a "solid material" to a differential potential will create a field who could cause something like electro migration. Depending of the field applied the particle in the matrix will stretch or contract the conductible material and create/destroy bridge. A movement of a few atom at that scale doesn't require a lot of power and in a matrix atom without a solid crystalline structure it would be made rather easy by the available hole in the structure.
For the number of cycling the material can do, they rate a rather big number of cycle. For that part I still held my doubt but for the rest they seem ok.