What Will Make Computers Faster: An Approach Based on Computational Complexity
Vladik Kreinovich
Traditional engineering approach to computer design is that we use processes which are well understood and easy to simulate – so that we can simulate how the resulting computational device will perform without the need to build and test an expensive hardware prototype. While in the past, this approach has led to the current great computer speed-up, we seem to be exhausting the this approach’s potential. New ideas are needed. In this paper, we argue that using well-understood but computationally-difficult-to-simulate processes is an alternative promising way to make computers faster. Indeed, the very fact that these processes are difficult to simulate means that if we simply wait for the result of this process, we get the results of the corresponding model faster than by simulating this process in a traditional computer. Thus, by incorporating difficult-to-simulate processes into computations, we can indeed makes computer faster.
Keywords: Future of computing, computational complexity, difficult-to-simulate processes