Bypassing the Silicon Speed Barrier
There is a fundamental physical barrier in the continuing improvement in computer speed. Silicon, the semiconductor material computer chips are made from, has a finite limit, a finite speed at which it can transmit electrical signals. This maxes out at around 100Ghz. However, long before that is reached, the inefficiencies in silicon, start converting most of the electrical energy to heat, meaning more and more powerful cooling systems are required, simply to keep going.
Researchers have long known of the problems, and we are expected to hit the final immovable barrier at around 2020.
However, long before that, as is happening now, the speed of computation will generate such excesses of heat that we will be forced to downgrade processing power or risk an energy consumption nightmare.
This is not a good thing for VR, as the processing power of a modern, average personal computer is not even a fraction of 1% of what is required for true immersion of our senses.
In order to achieve the processing power such applications demand, an alternative to silicon chips is required.
To do this, new semiconductor materials are needed.
A semiconductor can conduct electrons in one state, then be 'switched' to another state, where the same material has a very low conductivity, essentially turning the flow of electrons off. This switching between states on a single wafer is how circuitry is laid out across its surface.
Current commercial experiments are revolving around silicon germanium, which is a crossbreed material, an alloy basically. It contains a mix of both of silicon and germanium in unequal amounts. The germanium lowers silicon's natural electrical resistance, allowing current to flow with about half the heat emission, and upping the physical speed limit to just shy of 200Ghz.
International Business Machines (IBM) has already begun switching to silicon germanium technology, and other manufacturers are following. The problem to contend with, unfortunately is that silicon germanium (SiGe) still uses silicon, which slows maximum possible data transmission rates down, still generating heat every time the electron flow passes through a silicon rich component of the alloy.
As a result, SiGe processors typically extend battery life, or run cooler than their silicon brethren, but are not significantly swifter. If the speed was to increase, the heat dissipation issues of silicon would return, proportionally.
The only way we will achieve rates of information processing which go beyond mere gigahertz, is to find a new semiconductor material, one with next to no electrical resistance in its 'on' state.
Currently, Georgia Tech physics professor Walter de Heer is stunning the chip making world with his discovery that graphene - part of pencil lead - can be turned into a semiconductor.
Theoretical models had previously predicted that graphene, a form of carbon consisting of layers one atom thick, could be made into transistors more than a hundred times as fast as silicon transistors. In other words, computer circuits with a maximum threshold of 10,000 GHz, or 10 THz. This would be vastly more than the necessary threshold to completely subsume our senses into an artificial world, and create realities to rival meatspace.
Normally, graphene has a very, very high conductivity, but that can only be influenced very slightly - electricity will still flow through it very easily, and the flow cannot be turned 'off'. However, de Heer has used a computer model to show that if graphene could be fashioned into very narrow ribbons, it would begin to behave like a semiconductor.
Sadly, de Heer has not yet been able to make graphene ribbons narrow enough to behave as predicted. What he has done, is find a shortcut to achieving the same result.
By chemically modifying graphene with the addition of oxygen, a semiconductor material is created.
Heer has already used graphene to make transistors. In work performed at MIT, he was able to make several hundred graphene transistors, which worked together on a single, if very large, computer chip. This crude prototype demonstrated the concept was sound, and, once the scale is reduced, offer a route to eliminate the waste heat issues with computer chips without dropping their speed. On fact, with heat dissipation out of the way, chips made out of graphene instead of silicon, using the same lithography process, would be faster than their silicon brethren purely because that energy was going into data pathways and not heat generation.
As the final nail in the coffin, silicon cannot be carved into pieces smaller than about 10 nanometers without losing its attractive electronic properties. On the other hand, the basic physics of graphene remain the same in pieces smaller than a single nanometer.