GCC performance

I am doing parallel programming with MPI on Beowulf cluster. We wrote parallel algorithm for simulated annealing. It works fine. We expect 15 time faster execution than with serial code. But we did so