How to parallelize a program that need 0.005 seconds to terminate sequentially?
问题 for a project i need to parallelize a sudoku solver, that sequentially terminate in avarage in 0.005 seconds. For test i also count how many "computations" the program does to solve the sudoku, and the problem is that when i run it in multithreading, if i do a lot less "computations" of the sequential version, then the completion time of the program in multithreading is less than the sequential version, but if the sequential version does for instance 2000 computations to solve the sudoku, and