I am on C2D (Core 2 Duo Intel) and your code is working fine on my machine; all threads started and both core had used;

I make a little change; I renamed your class from "shared_mutex" to "shared_mutex1" and same for constructors and destructors.

I was getting redefination of that class. After changing name; every thing is working fine.

    io_service.stop();
    thread1.join();
    thread2.join();
    thread3.join();
    thread4.join();