New transistor designs (1) may help surpass the daunting constraints faced by contemporary software as we move toward solving highly specialized problems. The dominance of software over general computing architecture may be coming to an end. Until architectures can practically incorporate quantum computing, there is an impasse. And, quantum tunneling has brought conventional transistor designs to an asymptotic and predictable lethargy.
It is time to return to customized hardware for specialized problems. This will require more engineers again and less programmers and mathematicians. Educational institutions, grand masters at following the latest trends, while they teach the students "strategy," are incompetent in understanding what tomorrow will bring. And, that will certainly dampen innovation in the required direction.
However, this could be a short term phenomenon. Either of the two directions, squeezing more out of Silicon or finding better materials, may yield interesting designs that may allow a return to software in the future. The latter has much more potential but is also more risky. That would mean that the giants of the industry will shy away from it as they try to tie the quarterly numbers for the shareholders. And some of them may realize that the "cloud," is not a solution, just a sojourn.
The hardware regime is beginning again. We have to give electrons a more direct way to shuttle for efficient computing. Anything less will tie us down to the status-quo.
(1) http://advances.sciencemag.org/content/5/2/eaau7378
It is time to return to customized hardware for specialized problems. This will require more engineers again and less programmers and mathematicians. Educational institutions, grand masters at following the latest trends, while they teach the students "strategy," are incompetent in understanding what tomorrow will bring. And, that will certainly dampen innovation in the required direction.
However, this could be a short term phenomenon. Either of the two directions, squeezing more out of Silicon or finding better materials, may yield interesting designs that may allow a return to software in the future. The latter has much more potential but is also more risky. That would mean that the giants of the industry will shy away from it as they try to tie the quarterly numbers for the shareholders. And some of them may realize that the "cloud," is not a solution, just a sojourn.
The hardware regime is beginning again. We have to give electrons a more direct way to shuttle for efficient computing. Anything less will tie us down to the status-quo.
(1) http://advances.sciencemag.org/content/5/2/eaau7378