First off we will start with what a memristor is and promise this won’t make your eyes glaze over to bad, as this is one of the major disruptive technologies that is about to make its way on to the computing scene which will change it forever. In short this is a form of resistor as those familiar with electronics know the current triad of fundamental components as being the resistor, the capacitor, and the inductor. However in 1971, a University of California, Berkeley, engineer ( Leon Chua) predicted that there should be a fourth element to complete the mix. He predicted it would be a “memory resistor”, or “memristor” for short; however the technology in 1971 didn’t exist to build one.
Fast forward if you will some 37 years to 2008 and Hewlett-Packard researchers revealed that the solution was hiding in plain sight from them and technology had now caught up to theory allowing for the nano-scale construction of this device. Here we find the memristor is radically different from its sibling fundamental circuit elements as, unlike them it stores a memory of its past. Here, when you turn off the voltage to the circuit, the memristor will still remember how much was applied and for how long. This is an effect that can’t be duplicated by any circuit combination from the current triad of components comprised of resistors, capacitors, and inductors, and hence why the memristor qualifies as a fundamental circuit element in itself.
Ok, so what to do with this new marvel of mathematic wizardry and nano-scale engineering you ask yourself as you take a sip of your Starbucks coffee? Alright you better sit down for this one, as think of it this way, the new element will pave the way for applications from nonvolatile RAM, CPU’s to realistic neural networks. Oh and did I share with you it will do it at a fraction of the size and power of today’s devices? How small you ask, estimations of first gen devices will be petabytes in the same space terabytes with the possibility of Exabyte range in the future! The power, almost none as the memristor is a state-full device and unlike RAM or spinning storage (hard disks) its only needs a smidgen of power for read/write operations.
So how does this work to pack so much into such a small area you wonder? Well think of it this way, in a regular RAM chip you have an area of circuitry which makes up one bit, in computer terms a 1 or 0. Where in the very same space you will be able to “stack” many bits into one area as the memristor will be able to have more “values” then just a 1 or 0, in fact it will have tens of thousands! There is one catch, technology is working out how to maintain data integrity as in a binary operation it is pretty easy, however when you raise this to the thousands, or tens of thousands, the game changes as the substrate which the chip is made from including its environmental (temperature) considerations and check-sum controls become critical factors.
So bottom line, what does this all mean? In the short term, it will mean over packed datacenters will empty out as large storage frames will be replace with one a thousand times larger capacity wise, however physically it will only be the size of an ice chest. With the power consumption and cooling for the same will plummet as all those energy spinning platters head to the recycling heaps. This will also mean the rise of the “state-full” computer as the reason computers have to be booted every time they are turned on is that their logic circuits are not capable of holding their bits state after the power is shut off. However, because a memristor will remember it’s state, a memristor-driven computer would then never need to “reboot” or restart as it would always be on. Whit this one, hold on to your hats as HP has already signed up Hynix Semiconductor to start commercial production in 2013. Remember fact is always better then science fiction…