c++ - what is the fastest way to notify another thread that data is available? any alternativies to spinning? -

one thread writes data circular-buffer , thread need process data asap. thinking write such simple spin. pseudo-code!

    while (true) {         while (!a[i]) {             /* nothing - keep checking on , on */         }         // process b[i]         i++;         if (i >= max_length) {             = 0;         }     } 

above i'm using a indicate data stored in b available processing. probaly should set thread afinity such "hot" process. of course such spin expensive in terms of cpu it's ok me primary requirement latency.

the question - should write or boost or stl allows that:

1. easier use.
2. has same (or better?) latency @ same time occupying less cpu resources?

i think pattern general there should implementation somewhere.

upd seems question still complicated. let's consider case when need write items array in arbitrary order , thread should read them in right order items available, how that?

upd2

i'm adding test program demonstrate , how want achive. @ least on machine happens work. i'm using rand show can not use general queue , need use array-based structure:

#include "stdafx.h" #include <string> #include <boost/thread.hpp> #include "windows.h" // sleep   const int buffer_length = 10; int buffer[buffer_length]; short flags[buffer_length];  void processorthread() {     (int = 0; < buffer_length; i++) {         while (flags[i] == 0);         printf("item %i received, value = %i\n", i, buffer[i]);     } }   int _tmain(int argc, _tchar* argv[]) {     memset(flags, 0, sizeof(flags));     boost::thread processor = boost::thread(&processorthread);     (int = 0; < buffer_length * 10; i++) {         int x = rand() % buffer_length;         buffer[x] = x;         flags[x] = 1;         sleep(100);     }     processor.join();     return 0; } 

output:

item 0 received, value = 0 item 1 received, value = 1 item 2 received, value = 2 item 3 received, value = 3 item 4 received, value = 4 item 5 received, value = 5 item 6 received, value = 6 item 7 received, value = 7 item 8 received, value = 8 item 9 received, value = 9 

is program guaranteed work? how redesign it, using of existent structures boost/stl instead of array? possible rid of "spin" without affecting latency?

if consuming thread put sleep takes few microseconds wake up. process scheduler latency cannot avoid unless thread busy-spinning do. thread needs real-time fifo never put sleep when ready run exhausted time quantum.

so, there no alternative match latency of busy spinning.

(surprising using windows, best avoided if serious hft).