Microprocessor Commodities

Once in a while, I come across someone who wonders why I have such an interest in working with microprocessors. Often, these people would tell me that microprocessors are already a commodity and there is no value in working with microprocessors anymore. Whenever I hear this, I just smile and walk away. There is no reason to argue with people who do not understand microprocessors and what they actually do.

Microprocessors are definitely a commodity today, but only specific kinds of microprocessors with no applications – the venerable general purpose microprocessors. You can license, buy and get various architectures that will happily solve your problems until you actually try to solve some really cumbersome real-world problems and realise that your microprocessor lacks the computational power to do it.

There are still two diverging areas for microprocessors to evolve through.

One, microprocessors can evolve vertically to become more and more application specific. We already see that happening in the evolution of DSPs and other multi-media processors. However, these are basically specialised architectures of generic processors that turn them into much better number crunchers than commodity processors. In fact, most commodity processors have such ‘extensions’ built into them. Extrapolate this and you will see that as the market evolves, processors will also evolve to cater to more and more application specific uses.

Two, microprocessors can evolve horizontally to embrace more esoteric architectures in order to better model and solve real-world problems. In fact, commodity processors are very bad at solving a lot of real-world problems. Once you do anything more complicated than adding two numbers together, things start to fall apart. In fact, you can think of commodity processors as simple adding machines. Take the example of solving systems of partial differential equations, for example. Present day computer technology is particularly bad at solving these problems.

Furthermore, I am not even going into those computational realms of biological and quantum processors at all. Once those areas open up, things are going to get really hairy. So, anyone who thinks that microprocessors are dead, obviously do not know a thing about microprocessors. They are far from dead and there is still plenty of room to evolve and innovate.