The Miraculous Memristor - Logic And Memory Plus Going Beyond Moore's Law...
I recently visited HP Labs and spoke with Stan Williams, senior fellow at Hewlett-Packard and director of Quantum Science Research, about an incredible semiconductor device -- the memristor.
Until fairly recently, the memristor, short for memory resistor, was a mythical electronic component. It had been predicted to exist by a mathematician, Leon Chua, a professor at UC Berkeley, in an 1971 paper. No one had made a memristor until Stan Williams and his team cracked it in 2008.
- store data like DRAM or Flash but it doesn't require any energy to maintain the data storage.
- the memristor chips can be laid down in layer upon layer upon layer, creating three-dimensional structures that can store and process data.
- the memristor is easy to make and completely compatible with today's CMOS chip making processes.
- it can be scaled to very small geometries without losing its properties.
- the memristor can also perform logic, it can act as a microprocessor!
It's that quality of a memristor, that it can be used for both data storage and data processing that Mr WIlliams and his team recently discovered, and that potentially turns the world of computing on its head. This is a very big deal indeed.
Here are some notes from my conversation with Mr Williams:
- IT systems now take up 2 per cent of the world's electric power. We need IT systems that are 1,000 times more powerful, we need what is called exascale computing (The Need For A Radical New Type Of Computer Architecture). Yet to get there it would take too much electric power. The memristor makes exascale possible.
- My team first started looking at molecular electronics in 1997. We were seeing voltage curves that we couldn't understand, they had a signature figure of eight shape. One of our team came across a paper by Leon Chua that had a similar voltage curve. We had to scrap everything we were doing and focused on trying to create a memristor. We focused on titanium dioxide.
- We weren't getting very far so I decided to take the month of August off (8th month) and read everything I could about titanium dioxide, from geology, physics, chemistry, math. Titanium dioxide was studied separately by all these disciplines, it was known as a ceramic, a mineral, a dielectric, a semiconductor, even a sunscreen. By studying all of this separate research, on August 26, I finally figured it out, and brought all the pieces together.
- It will take us about 3 years before we have commercial memristors. The good news is that they are easy to make and completely compatible with standard CMOS chip manufacturing techniques. They also scale to small features very well. And standard chip design tools work very well for memristors.
- Memristors could save a lot of power in data processing because they don't require any power to maintain their data storage.
- HP is interested in memristors because we are the world's largest buyer of DRAMs.
- Memristors could be used to replace disk drives and DRAMs. They will initially be used in IT systems, in-between disk drives and DRAMs but then they can start to 'eat' into both sides -- disk drives and DRAMs.
- This gets interesting. We have now discovered that memristors can be used for logic -- they can be used as processors. This is very significant because instead of shuttling data to the processor and then back again, which takes time and energy, we could shuttle the processing code to the data -- which is smaller and quicker.
- I used to be worried that we are going to reach the limits of Moore's Law and the what do we do. But with memristors, we can easily lay down multiple layers of memristors, effectively extending Moore's Law by decades.
- Using memristors for processing brings other potential changes. Instead of just two states, on or off, as with transistors, memristors can represent many states. This means we can create new types of computing models, we can also create analog computers, which you don't program, but you let them learn. You can then replicate the learning to other memristor analog computers.
- We might be able to use memristors in a similar way to synapses in the human brain.
- We spoken to Intel about memristors many times. They seem to be more interested in other types of memory such as phase change memory.
- - -