同时使用多个线程c++

我在madelbrot程序中使用多线程时遇到问题。

我厌倦了学习教程

    int sliceSize = 800 / threads;
    double start = 0, end = 0;
    for (int i = 0; i < threads; i++)
    {
        start = i * sliceSize;
        end = ((1 + i) * sliceSize);

        thrd.push_back(thread(compute_mandelbrot, left, right, top, bottom, start, end));

    }

    for (int i = 0; i < threads; i++)
    {
        thrd[i].join();
    }
    thrd.clear();

但是,当使用8个线程时,代码只需要一半的计算时间。

我还尝试了更复杂的方法,但根本不起作用

void slicer(double left, double right, double top, double bottom)
{
    /*promise<int> prom;
    future<int> fut = prom.get_future();*/


    int test = -1;
    double start = 0, end = 0;
    const size_t nthreads = std::thread::hardware_concurrency(); //detect how many threads cpu has
    {
        int sliceSize = 800 / nthreads;

        std::cout << "CPU has " << nthreads << " threads" << std::endl;
        std::vector<std::thread> threads(nthreads);

        for (int t = 0; t < nthreads; t++)
        {

            threads[t] = std::thread(std::bind(
                [&]()
            {

                mutex2.lock();
                test++;

                start = (test) * sliceSize;
                end = ((test + 1) * sliceSize);

                mutex2.unlock();

                compute_mandelbrot(left, right, top, bottom, start, end);


            }));
        }
        std::for_each(threads.begin(), threads.end(), [](std::thread& x) {x.join(); }); //join threads
    }
}

但似乎当它同时计算8件事情时,即使在使用互斥锁后,它们也会跑得太快,而且速度也不会更快。

这让我头痛了7个小时,我想自杀。帮助