But Johnson notes that magnetism is already catching on in circuit design: some advanced devices are beginning to use a magnetic version of random access memory, a type of memory that has historically been built only with conventional transistors. “I think a shift is already under way,” he says.

This article is reproduced with permission from the magazine Nature. The article was first published on January 30, 2013.

5 Comments

1. 1. jtdwyer 06:16 PM 1/30/13

   This is interesting fundamental research, but the articles says:
   "The voltage approach is highly reliable and easy to miniaturize, but has its disadvantages. First, keeping the voltage on requires power, which drives up the energy consumption of the microchip. Second, transistors must be hard-wired into the chips and can’t be reconfigured, which means computers need dedicated circuitry for all their functions."
   
   1) Can the magnetic switch be continuously miniaturized as the voltage switch has been over these many decades?
   
   2) Power requirements are a fundamental issue for voltage based switching, but they are also reduced with miniaturization.
   
   3) Is there any real requirement or demand for dynamic hardware reconfiguration, or is this simply a capability of magnetic switches in search of a requirement, being portrayed here as a disadvantage for electronic devices? To the extent that there is any requirement for reconfiguration, hardware circuitry supports using reduced instruction sets that can emulate varying instruction sets, allowing a single chip architecture to perform various dedicated functions. 'Self-healing' capabilities similar to those described could be provided in electronic circuits by having unused circuit capacity that could be dynamically microprogrammed to recover some failed component for some hypercritical applications (processors on board satellites subject to cosmic radiation, for example), for most applications processor reliability is more than adequate. How many PCs suffer from processor failure?
   
   Again, interesting research, but at this point mostly a solution in search of a problem...

   Reply | Report Abuse | Link to this

2. 2. Autaut in reply to jtdwyer 07:08 PM 1/30/13

   Mostly true,
   
   However i might note, there are actually reconfigurable circuits available. they are called FPGA and come with two characteristics, that make runtime-rearranging unattractive:
   First, it takes a considerable amount of data to store the layout, which means "reprogramming" it takes time on the order of millis. This is days in processor time.
   On the other hand, they require a generic logic layout that leads to poor logic densities compared to ASICs (application specific IC).
   They are not widely used, because mass-produced ASICs are more powerfull and cost less.
   
   By the way, there is a flaw in the reasoning about power consumption:
   Today's ICs use MOSFET transistors which have an isolated gate. This means, holding the voltage takes only a very small amount of current. As we know, power is the product of both.
   Infact, very high current is needed to change the state of the transistors, even if only for a short time.
   This switching current accounts for the vast amount of power dissipation in all high frequency applications.
   (yes, that's the main reason why stepping down the clock on an idle cpu saves battery)
   
   This principle would also apply to such magnetic gates, as power is required to run through the magnetic hysteresis.

   Reply | Report Abuse | Link to this

3. 3. jtdwyer in reply to Autaut 07:33 PM 1/30/13

   Thanks for clarifying my potshots - I figured I'd hit something, with such a large target!

   Reply | Report Abuse | Link to this

4. 4. priddseren 10:00 PM 1/30/13

   While there could be some interesting if limited application of this technology, perhaps combined with the concept of medical nanobots that could be reprogrammed magnetically from outside the body it still suffers from the same basic problem all advances in computers have. The use of binary states for processing information. This is around 1940 tech and all that has occurred since then is making systems do this faster but still it is the same thing. The human brain doesn't use binary math and is able to achieve far more raw computing power than super computers, running on the energy of a carrot.
   The real advance in computers will be when someone invents a CPU that uses something more advanced than binary math.
   
   This problem occurs elsewhere. The 4 stroke internal combustion engine is a good example. Yeah, we have faster, better, more efficient engines today but in the end all we have done is taken an invention from the late 19th century and we just keep improving the devices doing it but we have not really come up with something more advanced. Or if there is something better, it has not displaced the over century old 4 stroke engine.
   
   These magnetic circuits are the same thing done a different way and not all that innovative from the perspective of computer processing. More of an innovation in electrical engineering.

   Reply | Report Abuse | Link to this

5. 5. komorkowy 11:38 AM 2/7/13

   Hello First of all, magnetic keeping the voltage on requires power on this website is explanation how does it work http://www.electrogsm.pl which drives up the energy consumption of the microchip. Second, transistors must be hard-wired into the chips and can’t be reconfigured

   Reply | Report Abuse | Link to this