In the CNET article “End of Moore’s Law: It’s not just about physics” the economic end of the doubling of computer power every year or two is predicted, because the R&D investment in nanometer (nm) scale production will not give a return on investment (ROI) sufficient to make it “good business”.
So, are we doomed, technologically speaking, because the power of our tablets won’t zoom ahead as fast as we’re used to? Or will the plateau be a good thing? After all, new cars are still being made, with only gradual improvements in performance. Old computers will wear out, needing replacement. Of course, a very different environment than we’re used to, but not Earth shattering.
The stable platform of hardware development would allow time for mature programming techniques to evolve to make better use of known technology, rather than always playing catch up, or being allowed the luxury of poor coding because the power of the CPU forgives inefficiencies.
This isn’t that different from the early days of the microprocessor, where gifted and talented coders made a lot out of a little. The return of the programming wizards will be presaged by increased understanding of how to use parallel architectures more effectively than we do now. So, I am not daunted by a slowing down of progress.
However, the article neglects to consider new technologies that aren’t just about cramming transistors onto silicon. Optical transistors, quantum computers, and even biologically grown processors – no silicon chip has come close to the capabilities of a human brain! – all promise the potential of lifting Moore’s Law by several orders of magnitude. But, and it’s a big but, it’s not the size, it’s what you do with it.