How to use parallelism to improve performance of my code? (tried a few things)

Question

So here is the problem: I have a vector of custom objects and I need to compute some stuff with each unique combination of objects in the vector, and if the result is a certain