I'm working on a class project that implements multithreading in a Sudoku Solver. My current solver works single threaded like this...
How the Solver works
Sets up the sudoku puzzle in 9x9 cells,starts in the top left of the puzzle and does 3 checks(row check, column check, box check), each cell has a possibility array the size of 0-8, if one of the checks finds a number in the check then it will set that cells corresponding array element to 1(this means that position is not a possibility for the answer of that box). When the array has 1 element left that is 0 then it fills in that cell with the answer.
My Question
I'm not trying to make the fastest sudoku solver but I just want to show the time difference in single threaded versus multithreaded. Which thread library would be best to use for this? What it be best to run 3 threads that each handle one of the checks and run it through multiple puzzles or would it be better to have multiple threads running their own puzzle and solving it. Is there a better way to show this that you can think of?
I would try one of these:
boost
openMP
or intel TBB
the easiest to use is openMP since you need to add some #pragma above your code.
boost or intel TBB means you really need to modify your code.
here are some links: http://openmp.org/wp/
https://www.threadingbuildingblocks.org/
http://www.boost.org/
please know that different CPU's and different implmentation will give you different result
you also need to you a profiler for better understanding where the optimizations would help you.
If you are running on Visual Studio you have a built in profiler for performance.
hope it helps to get started.