TABLE OF CONTENTS This chapter has presented an overview of the various mechanisms available for thread synchronization. Table 3.1 offers a recap of these synchronization mechanisms.
| Objective | Windows | Comments | Pthreads | Comments |
|---|---|---|---|---|
| Mutual exclusion | Mutex | Much slower than critical sections | Mutex | |
| Critical Section | More frequently used and faster than mutexes | N/A | ||
| Interlocked Function | High-speed integer manipulation | N/A | ||
| Synchronization | Spin wait | Fast but wasteful | Same as Windows | |
| Events | Cue threads off the actions of other threads | Condition variables | Suited to multiple threads | |
| Semaphore | Very rarely used | Semaphore | Commonly used. Best at controlling single threads |
We hasten to add that these discussions are just an introduction. Many complexities are intentionally not discussed because they're not germane to our overall goal. Entire books have been written on thread local storage, and the system cleanup of killed threads, among many other topics. For the most part, these issues arise only in complex parallel code. If you come to the point where you need this information, our recommendations for titles and resources that can guide you are located in the bibliography at the back of this book.
Two other topics discussed later are thread priority and thread affinity. Thread priority deals with which thread gets to run first when multiple threads can be scheduled. Thread affinity is the means by which a programmer tells the operating system on which processor to run a specific thread. Both of these...
