Magnetic Computing Takes a Step Forward
MaceyHW writes "PhysOrg.com reports a big step forward in the development of magnetic microchips. Since their initial creation of a magnetic logic gate in 2002, an international team of researchers from Durham University, Imperial College, London and the University of Sheffield 'team has created a number of further "logic gates" and created interconnecting structures using magnetic "nanowires," which can now reproduce the logic functions of a conventional computer.'"
I have, erm, sources connected to this story.
Basically, don't expect miracles. The signals can travel incredibly fast through these devices (can't remember exact figures) but there are problems. The signals are driven by an external magnetic field. Single magnetic pulse can drive high speed data transmission, but not normally operation of the circuit. Continuous operation would likely be in the 10s of kHZ to 1s of MHz range.
The main uses are in low power, low demand devices. Or in systems which need to be resistant to EM radiation (certain military interest there).
Similarities to core memory? Basically that's MRAM. And chips of MRAM up to about several hundred kb can now be mass produced.
But as for "they have made a computer" - we're not talking Pentiums. We're talking about having demonstrated the functions which compose such chips - NOT, AND, OR and signal combination and splitting. Nothing more.
Yes, it was rather vague. A quick search with google turned up a more detailed report on their earlier work:( I used "bug me not" to login to read it
Just to give the flavour of this report, it states...
Much of the funding has come from Eastgate, a Venture Capital firm based in Dubai. I think the IP rights are shared between them and the university, but Eastgate won't develop the idea. Rather, they'll look for people who are willing to.
IBM has been working on mram for some time now. It is magnetic ram. It is non-volatile memory. It does not lose its state even when power is removed. It could replace all existing memory(rom, flash, hard drive, ram).
The method referenced is called "magnetic domain-wall logic". It works by propagating domain walls in small ferromagnetic wires. They have demonstrated various logic elements. Unfortunately, the domain-wall propagation is powered by immersing the circuit into a rotating or alternating magnetic field. This requires bulky external coils, rotating the IC itself inside a permanent magnet, or rotating a permanent magnet around the IC.
In the demonstrations, field frequency is measured in the tens of Hertz. I could not find a reference to the propagation velocity, but one wire width per cycle would be a reasonable guess. In short, horribly slow.
The circuits are pretty much unaffected by most external magnetic fields. The problem with them is actually producing a field which a) is strong enough to affect them and b) can change fast enough to actually do useful data processing. The 200nm wide wires used (some down to 120nm) can only support one magnetic domain across them, so you can't 'corrupt' a single bit other than by flipping it. And the field required to flip it would need to be really strong - probably hundreds of Oersteds. That's why the logic devices they use are so clever - they don't flip bits by applying magnetic fields to them, but rather the nanowires make use of geometry to force the magnetic domains into unusual junctions where they get rotated. The best analogy is a car doing a three point turn - a NOT operation. It starts the movement in one direction, and comes out of it facing anothher.
How about 100 GB of storage capacity for the cost of a memory card ? Magnetic microchips used in cell phones could make them fully functional video cameras. In addition, the chips are non-volatile, so startup lag will become a not-so-fond memory. They use much less power than electronic chips. They can be made much smaller, possibly as small as a few atoms. The examples they have already fabricated "use no silicon and require no multilayer processing and so can be manufactured at very low cost on flexible substrates, while offering non-volatility, radiation hardness and several hundreds of MHz of bandwidth" . They're talking about plastic chips. Pretty impressive.
The technology, which is still being developed, can be classified as "nanotech" and is called "magnetic domain-wall logic" and is based on spintronics. Lots of folks are working on this because many believe that spintronics will allow for great advances in areas from quantum computing to DNA based molecular electronic devices. This particular development is important because it represents the first actual construction of logic gates, which are the basis of computing. So far the group has produced a "NOT gate" and a "11-stage serial shift register / digital frequency divider" in a 200nm design rule. They have also demonstrated the transfer of magnetic information without the use of magnetic fields. This paves the way for hybrid chips with both electronic and spintronic components. Such "3D chips" could contain many times the amount of information possible with current electronic chips. They will run cooler, with short "nanowire" pathways, and have the potential to surpass the performance of silicon chips. Moore's Law marches on.
billy - wonder if the "$100 laptop" guys have their phone number?