double sum_of_squares(T values)
{
// This is for 4 processors
// If you have a different number of processors,
// change the number of arguments to the parallel_invoke function accordingly.
Concurrency::combinable<double> sums;
Concurrency::parallel_invoke(
[&]
{ for(size_t i = 0 ; i<values.size() ; i += 4)
sums.local() += values[i]*values[i];
},
[&]
{ for(size_t i = 1 ; i<values.size() ; i += 4)
sums.local() += values[i]*values[i];
},
[&]
{ for(size_t i = 2 ; i<values.size() ; i += 4)
sums.local() += values[i]*values[i];
},
[&]
{ for(size_t i = 3 ; i<values.size() ; i += 4)
sums.local() += values[i]*values[i];
});
return sums.combine([](double lSum, double rSum){return lSum+rSum; });
}
unsigned long long factorial(int n)
{
Concurrency::combinable<unsigned long long> products
= Concurrency::combinable<unsigned long long>([]()->unsigned long long{return 1LL;});
Concurrency::parallel_for(1 , n+1,
[&products](int i)
{
products.local() *= i;
});
return products.combine([](unsigned long long lProduct, unsigned long long rProduct)
{ return lProduct*rProduct; });
}
int main()
{
Concurrency::combinable<double> piParts;
Concurrency::parallel_for(1, 1000000, [&piParts](long long n) {
piParts.local() += 6.0 / (n * n);
});
double pi2 = piParts.combine([](double left, double right) {
return left + right;
});
std::cout << "pi squared = " << std::setprecision(10) << pi2 << std::endl;
std::cout << "pi = " << std::sqrt(pi2) << std::endl;
return 0;
}