It will get very close. The delay is calculated after the work for "this" cycle has been done. The accuracy is limited by the accuracy of whatever you use to implement the delay, of course, and the system scheduler might preempt your process, but this can be limited by keeping the load reasonable. As I understand, a load average less than the number of processors means that all tasks are (on average) being run whenever they are not blocked.