I think that it would be a good Idea to realize a kind of thread scheduler that provides one (or two if hyperthreading is enabled) thread(s) per core. If a calculation is able to get parallelized, like the calculation of matrix determinants or the calculus of integrals, the thread scheduler should get...