OpenMp C++ algorithms for min, max, median, average [closed]

Answer 1

OpenMP doesn't support these reduction operations. Consider Intel Threading Building Blocks' parallel_reduce algorithm, where you can implement arbitrary algorithm.

Here an example. It uses summation of partial results. You may implement any function you want.

#include 
#include 
#include 
#include 


///////////////////////////////////////////////////////////////////////////////


class PiCalculation
{
private:
    long num_steps;
    double step;

public:

    // Pi partial value
    double pi;

    // Calculate partial value
    void operator () (const tbb::blocked_range &r) 
    {
        double sum = 0.0;

        long end = r.end();

        for (int i = r.begin(); i != end; i++)
        {
            double x = (i + 0.5) * step;
            sum += 4.0/(1.0 + x * x);
        }

        pi += sum * step;
    }

    // Combine results. Here you can implement any functions
    void join(PiCalculation &p)
    {
        pi += p.pi;
    }

    PiCalculation(PiCalculation &p, tbb::split)
    {
        pi = 0.0;
        num_steps = p.num_steps;
        step = p.step;
    }

    PiCalculation(long steps)
    {
        pi = 0.0;
        num_steps = steps;
        step = 1./(double)num_steps;
    }
};


///////////////////////////////////////////////////////////////////////////////


int main()
{
    tbb::task_scheduler_init init;

    const long steps = 100000000;

    PiCalculation pi(steps);

    tbb::parallel_reduce(tbb::blocked_range(0, steps, 1000000), pi);

    printf ("Pi is %3.20f\n", pi.pi);
}

Please check this link for additional reduction algorithms. http://cache-www.intel.com/cd/00/00/30/11/301132_301132.pdf#page=19 Please look through paragraph 3.3.1. There is an example on finding minimum value in an array.