Question

I’m trying to parallelize a dynamic programming algorithm in C++11
using this approach:

void buildBaseCases() {
        cout << "Building base cases" << endl;
            for (unsigned int i = 0; i < BOARD_SIZE; ++i)
            {
                buildBaseCase(i);
            }
        cout << "Done building base cases" << endl;
    }

So, my parallelized version would be something along the lines of:

void buildBaseCases() {
        cout << "Building base cases" << endl;
        #pragma omp parallel
        {
            #pragma omp for
            for (unsigned int i = 0; i < BOARD_SIZE; ++i)
            {
                buildBaseCase(i);
            }
        }
        cout << "Done building base cases" << endl;
    }

However, this is causing valgrind to complain about memory leaking.
Am I misunderstanding the way you’re supposed to use openMP, or is there something fishy going on?

Answer 1

It turns out there was no problem at all. This was a duplicate of the issue described here:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36298