I was recently pulling together a presentation on memristorsand the breadth of HP labs research activities. As I was looking at the material to cover, I noticed a great deal of research taking place on memristors outside of HP.
I do believe (contrary to what some may say) that is because this technology will fundamentally change the IT industry going forward.
At their heart, Memristors are a non-volatiles switch with great possibilities.
- Scales to just a few nanometers lateral size
- Multiple layers can be stacked for even greater total density
- Switching speed is <10 ns at a few picoJoules energy dissipation
- Have a state lifetime (value holding ability) for at least a few decades
The opportunities where they can be applied are:
– As the replacement for storage technology like: Flash, Solid-State and Magnetic hard drives and even dynamic random access memory
– Possibly even replacing many of the functions of custom integrated circuit and field programmable gate arrays.
The research on this area is still underway but the following numbers I was able to gather from a variety of sources, so they are presentative of the difference in capability between flash memory that everyone is so excited about and the memristor based equivalent.
Switching element size
Switching time (ns)
Set voltage (V)
Reset voltage (V)
-3 to -5
10^2 to 10^7
10^5 to 10^10
So in the coming few years as this is productionized, it will shift our expectations. For example, how would cloud be implemented differently if there was a terabyte of fast, static memory on our mobile devices? What if it were a petabyte??