A demo show for parallem_sum of hyper_sum by using of std::async.

parallel_sum.cc

#include <iostream>
#include <future>

template <typename IntT, typename UnaryFunc>
IntT do_hyper_sum(const IntT begin, const IntT end, const UnaryFunc func) {
    IntT res = 0;
    for (IntT i = begin; i != end; ++i) {
        res += func(i);
    }
    return std::move(res);
}

template <typename IntT, typename UnaryFunc>
IntT do_hyper_sum_parallel(const IntT begin, const IntT end, const UnaryFunc func) {
    static const constexpr size_t kParallelThrd = 100;
    const size_t diff = end - begin;
    if (diff <= kParallelThrd) {
        return do_hyper_sum(begin, end, func);
    }
    const IntT mid = begin + diff / 2;
    auto handler   = std::async(std::launch::async, do_hyper_sum_parallel<IntT, UnaryFunc>, mid, end, func);
    const IntT res = do_hyper_sum_parallel(begin, mid, func);
    return std::move(res + handler.get());
}

template <typename IntT, typename UnaryFunc>
bool hyper_sum(const IntT begin, const IntT end, const UnaryFunc func, IntT* const res) {
    if (not(begin < end)) { return false; }
    if (nullptr == res) { return false; }
    *res = do_hyper_sum_parallel(begin, end, func);
    return true;
}

int main() {
    using IntT = long long int;
    IntT begin = 1;
    IntT end   = 1000;
    IntT res   = 0;
    auto func  = [](const IntT num){ return num * num + num * num * num; };
    if (hyper_sum(begin, end, func, &res)) {
        std::cout << res << std::endl;
    }

    return 0;
}

On *nix, please make sure that you compile this snippet of code by

g++ -std=c++11 parallel_sum.cc -lpthread