Linux: This Book Includes 4 Manuscripts. The Underground Bible to the unix operating System with Tools On Security and Kali Hacking to Understand Computer Programming, Data Science and Command Line

that one could not know who signed up first (this name indicates that for
such an algorithm all flows are equal).
Each stream allocated time period in which this stream is allowed to run is
called a time slice ( or time quantum ). When the thread is still running
after the time quantum is exhausted, it is interrupted and the processor
switches to perform instructions from another thread. When it blocks or
completes its execution until the time quantum runs out, the processor is
also passed to another thread. The length of time quantum for the whole
system is the same.
Such an algorithm is quite easy to implement. For this, the turn of the ready
threads should be a cyclic list. When a stream has exhausted its quantum of
time, it is moved to the end of the list, where new flows are
provided. Checking the exhaustion of the quantum of time is performed in
the interrupt handler from the system timer .
The only characteristic that affects the operation of the algorithm is the
length of the quantum of time. The balance between time spent switching
contexts and having to respond to many simultaneous interactive queries
should be kept in mind here.
Tasking too short a quantum of time leads to a lot of context switches, and a
significant percentage of CPU time is spent not on useful work, but on
those switches. On the other hand, the task of a too-long quantum, while
saving CPU time, also reduces the response time for interactive queries
because if ten users press a key at the same time, ten threads will be in the
ready list, resulting in the last one expecting ten long quanta time. In the
case of a quantum of infinite length, circular scheduling is reduced to
the FIFO algorithm (all threads have time to block or end before the
quantum is exhausted). In practice, it is recommended to set the quantum
length to 10 - 100 ms.
Note that traditional circular scheduling can "skew" toward streams that are
constrained by processor capabilities. Such threads preferably utilize their
quantum completely, whereas threads limited by I / O often transmit control
until the quantum runs out, and as a result, they have less processing

time. To solve this problem, you can increase the length of the quantum
(given the problems described earlier) or enter an additional queue of
completed I / O streams, which takes precedence over completed queues .

Download 6,34 Mb.