dracodraconis: (Default)

Computer components shrinking faster than predicted - tech - 20 February 2009 - New Scientist

The New Scientist article summarizes two recent discoveries. The first is a novel form of magnetic memory that could store 10 terabytes per square inch, a significant increase over the current limit of 0.5 Terabytes per square inch. The researchers believe this approach could soon lead to memories with densities approaching 100 Terabytes per square inch. The second discovery is a new form of transistor based on perovskite, rather than silicon, that can be as small as 2 nanometers, far smaller than the 45 nanometres of contemporary transistors. The circuits can also be redrawn up to 100 times, making the circuit somewhat hardware rewritable.
dracodraconis: (Default)

the physics arXiv blog » Blog Archive » Memristors made into low cost, high density RRAM (Resistive Random Access Memory)

University of California researchers have determined that low-cost, high-density memristor-based memory is feasible with current technology. Current research is focused on figuring out which substances work best.

dracodraconis: (Default)
Researchers at Hewlett-Packard claim to have discovered the long-fabled memristor (memory resistor), completing the basic electrical toolbox of resistor, capacitor, and inductor. It's existence was first postulated in the 1970s but nobody had ever observed one, or had reason to try to make one, until now.

They created the memristor by doping titanium dioxide, resulting in a device in which the dopant migrates in the direction of the current. Apply a voltage in one direction and the dopant migrates into the pure titanium dioxide, lowering the resistance. Reverse the polarity and the dopant begins to migrate back, increasing the resistance. Most importantly, when power is removed, the circuit "remembers" the previously-applied voltage, encoded by the dopant, as well as how long the voltage was applied. This has potential application for emulating neural synapses, making it possible to build dense neural networks on-chip. They are now working to find out how to integrate titanium dioxide components with silicon-bsed devices.

January 2010

S M T W T F S
     1 2
3456789
10111213141516
17181920212223
24252627282930
31      

Syndicate

RSS Atom

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags