Friday, November 28, 2008

I Memristor You

Big changes could be coming to your computer in the very near future. The change I am looking forward to the most is instant on computing. Imagine pushing a button and seconds later just like your TV or radio your computer is ready to use.

Earlier this year scientists at HP have demonstrated a workable switching memristor. First conceived by Leon Chua in a 1971, the theoretical memristor was an idea waiting for material sciences to catch up. He believed that a fourth device was required to provide conceptual symmetry with the resistor, inductor, and capacitor. The memristor is needed to complete the basic passive circuit elements as defined by a relation between two of the four fundamental circuit variables, namely voltage, current, charge and flux. The venerable resistor is such a passive element as it relates to voltage and current giving us V = R × I or I = V / R or R = V / I, known as the formulas of Ohm's Law. The once theoretical memristor fills this role in the capacitance/inductance/resistive triad.

What this means in practical terms for the continuation of Moore's law - that shrinking transistor sizes, and by extention computing power would double approximately every two years - is that far fewer switching memristors could replace many transistors on a given circuit thus increasing the speed and density of the circuit. Given that fact that memristors can "remember" their current state even when power is removed is really the key to their potential.

In theory memristors can be combined into devices called crossbar latches, which would replace many of the transistors in our future computers. With circuits taking up a much smaller area computers can be smaller and more powerful than ever. The exciting concept for everyday computer users is that memristors can also be fashioned into non-volatile solid-state memory, which would create far greater data density than on existing hard drive technology with access speeds similar to DRAM. This could spell the end of hard drives, the single biggest source of failure in current computer architecture. HP has reported that its version of the memristor is about one-tenth the speed of DRAM - for now...

Increasing computer performance has usually meant shrinking components so that more transistors can be packed onto a circuit. However, transistors can't get much smaller. Instead, this breakthrough will allow the removal of some transistors to be replaced by a smaller number of memristors. The only thing keeping Moore's Law afloat right now since CPU speeds have seemingly reached their limit is stacking multiple cores per CPU.

The next few years will be very important for memristor research. Right now, "the biggest impediment to getting memristors in the marketplace is having so few people who can actually design circuits using memristors," Stan Williams, a senior research fellow at HP says. Still, he boldly predicts that memristors will arrive in commercial circuits within three years.

We have already seen the power of miniturization - just look at the computing power of a cell phone. Can implantable computers for our brains be far behind with developments like these? I for one could use a little more non-volitile RAM upstairs - if you know what I mean. I aint getting any younger.



meika said...

You're linked like a synapse in my story Carry Me Home.