For ease of understanding, express the computational effort of task in terms of the largest factorial (n!) one can compute in the same amount of computation time. This gives a nice round number, exponentially scaled to keep up with Moore's Law.
It's not the best way; computation of factorial depends on the algorithm, amount of memory available. It's a very rough order of magnitude comparison.
Inspired by Watson; just how much computation does Watson do per question? Of course it's all parallelized.
No comments :
Post a Comment