Hi Gaz, Yes, timing loops (busy loops) may be optimized out by GCC. (I'm laughing just thinking about -funroll-all-loops a busy loop.) Timing loops that use sleep(), usleep(), nanosleep() or wait() or some event-based execution suspension won't (can't) be optimized out by GCC. Timing (busy) loops that poll a volatile, high-precision volatile timer won't (can't) be optimized out by GCC. HTH, --Eljay
