FastMutex.cpp

#include <cstdint>
#include <Windows.h>
#include <cassert>
#include <thread>
#include <iostream>

struct Mutex
{
public:
    void lock()
    {
        const auto currentId = ::GetCurrentThreadId();
        unsigned long expectedOwner = 0;
        while (true)
        {
            if (::InterlockedCompareExchange(&owner_, currentId, 0) != 0)
            {
                ::WaitOnAddress(&owner_, &expectedOwner, sizeof(unsigned long), INFINITE);
            }
            else break;
        }
    }

    void unlock()
    {
        const auto currentId = ::GetCurrentThreadId();
        assert(owner_ == currentId);
        InterlockedExchange(&owner_, 0);
        WakeByAddressSingle(&owner_);
    }

private:
    unsigned long owner_{};
};

Mutex lock;

void Output()
{
    for (int i = 0; i < 10000; ++i)
    {
        lock.lock();
        std::cout << i << ' ';
        std::cout.flush();
        lock.unlock();
    }
}

int main()
{
    std::thread th1{ Output };
    std::thread th2{ Output };
    th1.detach();
    th2.join();
}