How to make my code dealing with calculations based on data from large data sets faster in C++?

Question

I am writing a code in which I have input data of size 50000x20 matrix (50000 blocks , 20 properties each). I then calculations to decide the block I would select to move based on some calculations for which I have to run the loop over all blocks. Then I make the some changes to a characteristic value (a different variable )for this block and repeat the same process over and over. Is do the calculations for each block inside the loop. Is calling a function a better and faster way to do it? This code is running without converging (based on some convergence criteria) for long. How so I deal with this? Is this approach of dealing with large data sets good. Can I do better? Please help.

Thanks Surendra

Answer 1

Calling a function helps with readability and can potentially even slow down your code if it's called extensively and is not inline .

You need to do benchmarking and determine what your bottlenecks are and also take a look at dynamic programming , it could be of help to you.

Would also help to post some code :).

Answer 2

Does changing the characteristic value for block affect the other blocks?

If not, then there's probably no need to repeat the calculations for the other blocks.