c++ - boost::threadpool::pool vs.boost::thread_group -

i'm trying understand different use cases. , difference between 2 thread uses. this great tutorial have read explains boost::thread_group.

and here code i'm using:

boost::threadpool::pool s_threadpool(getcorecount());  cfiltertask task(pfilter,  // filter run     boost::bind(&cfiltermanagerthread::oncompletetask, this, _1, _2) // oncomplete sync callback          // _1 filter name  // _2 error code                 );  // schedule new task - runs on threadpool s_threadpool.schedule(task); 

this destructor:

s_threadpool.wait(0); 

can please explain?

boost::thread_group convenience class performing thread management operations on collection of threads. example, instead of having iterate on std::vector<boost::thread>, invoking join() on each thread, thread_group provides convenient join_all() member function.

with boost::thread, regardless of being managed boost::thread_group, lifetime of thread dependent on work in thread doing. example, if thread created perform computationally expensive calculation, thread can exit once result has been calculated. if work short-lived, overhead of creating , destroying threads can affect performance.

on other hand, threadpool pattern, number of threads services number of task/work. lifetime of thread not directly associated lifetime of task. continue previous example, application schedule computationally expensive calculation run within thread pool. work queued within threadpool, , 1 of threadpool's threads selected perform work. once calculation has completed, thread goes waiting more work scheduled threadpool.

as shown in threadpool example, threadpool can implemented boost::thread_group manage lifetime of threads, , boost::asio::io_service task/work dispatching.