And then it would take a year or two to get to the point where the technology is rugged and reliable enough to be used in this application that I was talking about in an enterprise computing system as the buffer. Then, another year or two after that we would think about broadening it out and starting to replace DRAM and hard disk.
And at least a year or two after that we would start thinking about doing processing within the sea of storage. So thats a pretty ambitious timeline. If you dont set ambitious goals for yourself, youre sort of doomed to a slow path. Those are a set of research goals that we have.
Do you also believe that the memristor-base processors could potentially replace silicon chips across the world on a large…?
Well, memristors cant replace transistors, so in fact the chips that were talking about are hybrid chips. They contain both transistors and memristors but the hybrid circuit with the two together is much more powerful than a circuit with only transistors. So instead of replacements, think about this: the memristors will augment or supercharge the silicon and provide an underlying silicon layer with a vastly greater amount of capability than current chips have. Think about a silicon chip still with a layer of CMOS, but on top of that, multiple layers of memristors that can be used for data storage and local computation with the silicon CMOS acting as the power supply and the input-output devices for the memristors circuitry.
Tell us about any partnerships with other tech companies or enterprise customers you may have struck to develop this technology and further its commercial development.
Well, the actual research has actually been done completely internally, so the research part of this was independent and all done here at HP Labs in Palo Alto. Now that were going into the lab to fab phase, we are in the process of partnering developmental commercialisation partners and were not ready to announce the specific identity of those partners yet.
When do you see memristors being most widely used in handhelds, mobile computers…?
In three years, it will be handhelds, PCs, notebooks and slates. That would really be very nice because what that would mean for these devices is that there will be noespecially for the slates and notebooksboot-up time. You turn something on and it will be on instantaneously like a light bulb, so you dont have to worry about this issue of booting-up and booting-down because all of the storage will be random access and it will be non-volatile.
Within the five-year timeframe, we see memristor making inroads as solid state storage as this buffer between memory and hard disks in the enterprise space. Then in the seven-year time frame, we will see the start of the process of moving outwards, starting to flatten the hierarchy by essentially crunching on the DRAM space and the hard disk space with memristor technology.
Sign up for Computerworld eNewsletters.