Question

I’ve started using some basic threading in c++0x to speed things up. For a trivial example which does nothing except test speed:

void my_thread_func(int n, int steps, int offset)
{
  double a[n];
  for (int i=0; i<n; i++)
    a[i] = i + offset;
   for (int step=0; step<steps; step++)
     for (int i=0; i<n; i++)
       a[i] = sin(a[i]);
}

int main()
{
   std::vector<std::thread> t;
   for (int i=0; i<numRuns; i++)
     t.push_back(std::thread(my_thread_func, n, steps, i));
   for (int i=0; i<numRuns; i++)
     t[i].join();
}

and that works quite well. My question is somewhat a general one, namely how to generalize the above idea to working with member functions. Say I have 10 Rope(s), and each rope has several (single-threaded) member functions, computeTorques(), computeForces(), etc. Assuming the Rope(s) do not share memory/resources, is there a way I can multi-thread calling each member function of each Rope without explicitly having to change the (single-threaded) code within the Rope?

Answer 1

Use std::bind.

class Rope
{
    ...
    void compute();
};

int main()
{
   Rope ropes[10];
   std::vector<std::thread> t;

   for(auto& r : ropes)
      t.push_back(std::thread(std::bind(&Rope::compute, &r)));

   for(auto& th : t)
     th.join();
}

EDITED: Using lambda would be better.

   for(auto& r : ropes)
      t.push_back(std::thread([&]{ r.compute(); }));