Parallel computing in Java for big data

Parallel computing in Java for big data

There are two sets of matrices i.e. A(1 500 matrices) and B(150 000
matrices).
For each matrix a from set A I iterate throw all elements b from B
checking criteria on a and b (if it's true for some matrix b I stop
iterating and return true if not keep iterating till the last element of
the B set is reached and then return false).
My question is how to make this program parallel using Java?
I understand that the matrices sets are quite big (the numbers of elements
I posted above is just for instance, in real-life case they would be
bigger). How to correctly do all that computations? Create 1 5000 threads
and do all that computation is actually not a good idea. What is an
optimal amount of threads, which will take all computational work?