"Magnetic Tornadoes" Could Offer New Data Storage Tech 109
coondoggie writes to tell us about the latest technique researchers are investigating as a possible means to store data, magnetic tornadoes. "Conventional computer memories store data in "bits" that consist of two magnetic elements that record data in binary form. When these elements are magnetized in the same direction, the computer reads the bit as a '0'; when magnetized in opposite directions, the bit represents a '1,' researchers stated. According to scientists, a vortex forms spontaneously — one vortex per disk — in a small magnetic disk when the disk's diameter falls below a certain limit. Although the vortex does not whirl about like a meteorological tornado, the atoms in the material do orient themselves so that their magnetic states, or 'moments,' point either clockwise or counterclockwise around the disk's surface. At the center of the disk, the density of this rotation causes the polarity of the vortex core to point either up out of the disk or down like a tornado's funnel, researchers stated. Because the vortices that form on the disks contain two independently controllable and accessible magnetic parameters, they could form the basis for quaternary bits that would contain data written as a 0, 1, 2, or 3."
"Quaternary bits"? (Score:4, Informative)
Is a "quaternary bit" a "quaternary binary digit"? Doesn't make sense. I think you're after a "quaternary digit", or "quit".
Re:The end of binary (Score:4, Informative)
00b=0q
01b=1q
10b=2q
11b=3q
Each quaternary bit would store two binary bits, all translated by the device. Bytes would still be 8 binary bits, but only 4 quaternary bits. Much easier than translating between trinary and binary...
And, as they are talking about storage medium, NOT processors, there's no need to recompile. Just have the device handle the translation, much in the same way it's done for CDs and flash memory.
Re:There are 1 types of people who understand quin (Score:2, Informative)
It's just bit packing. For example (and ignoring many low level details*) your 512-byte sector would be stored in 2048 hardware bit buckets instead of 4096 individual storage quanta.
* For purposes of illustration and ignoring the smart little tricks of hardware reality.
Re:Re-discovering magnetic bubble memory (Score:4, Informative)
This sounds a lot like magnetic bubble memory [wikipedia.org] that intel, fujitsu, IMB and TI made in the 1980s.
That too had multiple states per "bubble". However the higher-order bubbles were generally not used. The reason was, it was hard enough keeping the single bit (zeroth order mode) bubbles stable at high circulation and high density.
Since here the domains are fixed and the disk moves it might be easier to use higher order magnetic domain modes.
Magnetic vortices are significantly smaller than the bubbles in bubble memory. Because of this, there are no "higher order" states - you have 4 distinct magnetization states (CW/CCW, in/out), and there are no in-between states. The trick is figuring out how to get the switching speed down using exchange bias coupling and crazy anisotropy effects.
Re:Information Theory? (Score:1, Informative)
I remember seeing this sort of thing discussed a long time ago. The thinking was to find themost economical way of storing/writing numbers, eg what is the most efficient base to use. Base 2 only needs two characters, but needs a long nubmer to store much data. Base 10 needs more characters but the nubmers end up much shorter. So which is best is going to depend on the relative cost of adding more places in the number or more characters in the storage set. If you make the assumption that either has the same cost, then e would be the optimum base to use. This does of course ignore the fact that a noninteger base is totally inconvenient to use, especially for integer numbers. Practically we are stuck with integer bases, so under the above assumption 3 would be the most economical. However, in real life the assumption does not hold true. For human counting it turns out to be easy to have more characters, so we have settled on base ten. We need more characters, 10 to be precise, but the numbers end up very short. Meanwhile, for machine based systems, it turns out to be generally simpler to stick to two characters (on, off) and simply add more places to get numbers of the desired length...8, 16, 32, and now 64 bits. (nothing stopping us having other lengths but powers of two are of course convenient.) If we did go to a three state logic, the circuits would become much more complex, and in practice two two-state circuits would be simpler than one three state.
Re:And a 1, 2, 3? (Score:3, Informative)
That would be four quaternary bits to make a byte, I believe.
2^8 = 256 possible values (binary; 8 places, 2 possible values each)
4^4 = 256 possible values (quaternary; 4 places, 4 possible values each)