Princess Nora University

Computer Sciences Department

Operating Systems

CS 340 T

1 Level 3 (IT) + Level 4 (IS) + Level 5 (CS)

 Chapter-4
CPU Scheduling

Introduction
Chapter 4: CPU Scheduling

1. Basic Concepts
2. Scheduling Criteria
3. Scheduling Algorithms

3
OBJECTIVES:

 To introduce CPU scheduling, which is the basis for

multiprogrammed operating systems

 To describe various CPU-scheduling algorithms

4
Basic Concepts

5
Basic Concepts

CPU–I/O Burst Cycle :

o Process execution consists of a cycle of
CPU execution and I/O wait
o Process execution begins with a CPU
burst…That is followed by an I/O burst,
which is followed by another CPU burst,
then another I/O burst, and so on.
o Eventually, the final CPU burst ends with
a system request to terminate execution

6
Histogram of CPU-burst Times

 The durations of CPU bursts

vary greatly from process to
process.
 There is large number of short
CPU bursts and a small number
of long CPU bursts.
 An I/O-bound program >>>>
has many short CPU bursts.
 A CPU-bound program >>>>
has a few long CPU bursts

7
 CPU Scheduler

 CPU Scheduler (/ short-term scheduler):

o Selects one process from among multiple ready processes

in memory, and allocates the CPU to one of them

o The ready queue is NOT necessarily a first-in, first-out

(FIFO) queue. It can be implemented as a FIFO queue, a
priority queue, a tree, or an unordered linked list.

8
Preemptive Scheduling

CPU scheduling decisions may take place when a process:

1. Switches from running to waiting state

2. Switches from running to ready state
3. Switches from waiting to ready
4. Terminates

 Scheduling under case(1) and (4) is non-preemptive

(/cooperative).
 scheduling under case (2) and (3) is preemptive

9
Preemptive Scheduling (cont..)

 What is non-preemptive scheduling?

Once the CPU has been allocated to a process, the
process keeps the CPU until it releases the CPU either
by terminating or by switching to the waiting
state.
 What is preemptive scheduling?
The process can be removed forcibly from the CPU
before finish executing because a better process arrives
to the ready queue.

10
Dispatcher

 Dispatcher : is an OS module gives control of the CPU to

the process selected by the short-term scheduler.

 The dispatcher should be as fast as possible, since it is invoked

during every process switch.

 Dispatch latency : is amount of time it takes for the

dispatcher to stop one process and start another running.
( i.e. context switch time)

11
Scheduling Criteria

12
Scheduling Criteria

1. CPU utilization
2. Throughput
3. Turnaround time
4. Waiting time
5. Response time

13
 Scheduling Criteria

 CPU utilization: the time percentage that the CPU is busy (/not
idle).

 Throughput : number of processes that complete their execution

per time unit (e.g. 10 processes /sec)

 Turnaround time: time interval from the time of submission of a

process to the time of completion
Turnaround time = waiting time to get into RAM + waiting time
in RAM + CPU execution time+ I/O time
Turnaround time = Termination time- creation time
14
Scheduling Criteria (cont.)

 Waiting time – amount of time a process has been waiting in

the ready queue .

 Response time – amount of time it takes from when a

request was submitted until the first response is produced.
(i.e. is the time it takes to start responding, not the time it takes
to output the response)
 In an interactive system, response time is very important.

15
Scheduling Criteria (cont.)

 It is desirable to:
 Max CPU utilization
 Max throughput
 Min turnaround time
 Min waiting time
 Min response time

16
Scheduling Algorithms

17
Scheduling Algorithms

1. First-Come, First-Served Scheduling (FCFS)

2. Shortest-Job-First Scheduling (SJF)
3. Priority Scheduling
4. Round-Robin Scheduling (RR)
5. Multilevel Queue scheduling
6. Multilevel Feedback Queue

18
(1) First-Come, First-Served (FCFS)

 FCFS is the simplest CPU-scheduling algorithm

Basic Methodology:
 The process that requests the CPU first is allocated
first.
 FCFS algorithm is always non-preemptive

19
(1) First-Come, First-Served (FCFS)

 FCFS can be implemented using FIFO queue:

 When a process enters the ready queue, its PCB is linked
onto the tail of the queue. When the CPU is free, it is
allocated to the process at the head of the queue. The
running process is then removed from the queue.

 Gantt chart: is a bar chart that illustrates a particular schedule,

including the start and finish times of each of the processes.

20
 (1) First-Come, First-Served (FCFS) -cont..

 Example(1): Consider the following set of processes that arrive at time 0, with
the length of the CPU burst given in milliseconds
Suppose that the processes arrive in the order: P1 , P2 , P3
Draw the Gantt chart using (FCFS) algorithm and find the average waiting time.

Process Burst time (ms)

P1 24
P2 3
P3 3

 The Gantt Chart for the schedule is:

P1 P2 P3

0 24 27 30

21
(1) First-Come, First-Served (FCFS) -cont..
 Example(1)- cont.:
Waiting time for P1= W(P1) = (0-0)=0 ms
W(P2) = (24-0)=24 ms
W(P3) = (27-0)=27 ms
Average waiting time = (0+24+27)/3 = 17ms

 Waiting time = Σ waiting intervals

time in to CPU- arrival time (if it is the FIRST interval)

 waiting interval= time in to CPU- last time out of CPU

(if it is the NOT the first interval)

22
(1) First-Come, First-Served (FCFS) -cont..
Example(2): Consider the same previous set of processes arrive at time 0,with the length
of the CPU burst in milliseconds
Suppose that the processes arrive in the order: P2 , P3 , P1
Draw the Gantt chart using (FCFS) algorithm and find the average waiting time.

Process Burst time (ms)

P1 24
P2 3
P3 3

 The Gantt chart for the schedule is:

P2 P3 P1

0 3 6 30
 W(P1) = 6 ms ; W(P2) = 0 ms ; W(P3) = 3 ms
 Average waiting time: (6 + 0 + 3)/3 = 3 ms >>Much better than example (1)..WHY??
23
(1) First-Come, First-Served (FCFS) -cont..

 Convoy effect :
 It is problem appears in FCFS algorithm where short
processes wait for the one big process to get off the CPU.
 This effect results in lower CPU and device utilization

24
(1) First-Come, First-Served (FCFS) -cont..

 FCFS Pros. (++):

 Simplest algorithm

 FCFS Cons. (--):

 The average waiting time is generally not minimal and
affected by processes’ order.
 Lower CPU and device utilization because of convoy effect
 The average response time is generally bad (i.e. Not suitable
for time-shared systems)

25
(2) Shortest-Job-First (SJF) Scheduling

Basic Methodology:

 Associate with each process the length of its next CPU

burst.
 Use these lengths to select the process with the shortest
time
 If the next CPU bursts of two processes are the same,
FCFS scheduling is used to select the next process.
 Two schemes: Non-preemptive & Preemptive

26
(2) Shortest-Job-First (SJF) -cont..

Two schemes:

Non-preemptive SJF Preemptive SJF

once CPU given to the process it If a new process arrives with CPU burst
cannot be preempted until length less than remaining time of
completes its CPU burst current executing process, preempt.

27
(2) Shortest-Job-First (SJF) -cont..
 Example(3): Consider the following set of processes .
Draw the Gantt chart using (Non-preemptive SJF) algorithm and find the
average waiting time.
Process Arrival time Burst time (ms)
P1 0 7
P2 2 4
P3 4 1
P4 5 4

P1 P3 P2 P4

0 3 7 8 12 16

 W(P1) = 0 ms ; W(P2) = 6 ms ; W(P3) = 3 ms ; W(P4) = 7 ms

 Average waiting time=(0 + 6 + 3 + 7)/4 = 4 ms

28
(2) Shortest-Job-First (SJF) -cont..
 Example(4): Consider the following set of processes .
Draw the Gantt chart using (preemptive SJF) algorithm and find the average waiting
time.
Process Arrival time Burst time (ms)
P1 0 7
P2 2 4
P3 4 1
P4 5 4

P1 P2 P3 P2 P4 P1
0 2 4 5 7 11 16

 W(P1) =9 ms ; W(P2) = 1 ms ; W(P3) = 0 ms ; W(P4) = 2 ms

 Average waiting time = (9 + 1 + 0 +2)/4 = 3 ms

29
(2) Shortest-Job-First Scheduling-(cont..)

 SJF Pros. (++):

SJF is optimal – gives minimum average waiting time for a
given set of processes

 SJF Cons. (--):

The difficulty is knowing the length of the next CPU request
(some times this time is predicted).

30
(3) Priority Scheduling

Basic Methodology:

 A priority number (integer) is associated with each process.

The CPU is allocated to the process with the highest
priority.
 Equal-priority processes are scheduled in FCFS order.
 Text book assumes (smallest integer  highest priority)
 Two schemes: Non-preemptive & Preemptive

31
(3) Priority Scheduling (cont..)

 Two schemes:

Non-preemptive priority algorithm Preemptive priority algorithm

Once CPU given to the process it If a new process arrives with priority
cannot be preempted until higher of current executing process,
completes its CPU burst preempt

32
(3) Priority Scheduling (cont..)
 Example(5): Consider the following set of processes .
Draw the Gantt chart using (Non-preemptive priority) algorithm and find the
average waiting time where higher priority in the system is (1).

Process Arrival time Burst time (ms) Priority

P1 3 6 5
P2 5 2 3
P3 6 1 2

P1 P3 P2
3 9 10 12

 W(P1) =0 ms ; W(P2) = 5 ms ; W(P3) = 3 ms

 Average waiting time = (0 + 5 + 3)/3 = 2.7 ms

33
(3) Priority Scheduling (cont..)
 Example(6): Consider the following set of processes .
Draw the Gantt chart using (preemptive priority) algorithm and find the
average waiting time , where higher priority in the system is (1).

Process Arrival time Burst time Priority

P1 5 6 8
P2 7 2 1
P3 9 3 8
P4 11 4 9
P5 25 1 3

P1 P2 P1 P3 P4 idle P5
5 7 9 13 16 20 25 26

 W(P1) =2 ms ; W(P2) = 0 ms ; W(P3) = 4 ms ; W(P4) = 5 ms ; W(P5) = 0 ms

 Average waiting time = (2 + 0 + 4 +5 + 0)/5 = 2.2 ms
34
(3) Priority Scheduling (cont..)

 Priority scheduling Pros. (++):

 Simple algorithm

 Priority scheduling Cons. (--):

 Main Problem : Starvation ( /indefinite blocking) where
low priority processes may never execute.

 Solution : Aging >>as time progresses, system increase the

priority of the process .

35
(4) Round Robin Scheduling (RR)

Basic Methodology:
 Each process gets a small unit of CPU time (time quantum).
 After this time has elapsed, the process is preempted and
added to the end of the ready queue.
 The ready queue is treated as a circular queue and
implemented as FIFO queue
 RR algorithm is always preemptive

 (RR) algorithm is designed especially for time-sharing systems (to improve

response time ) .

 Time quantum (/ time slice ) (q)>> usually 10-100 ms.

36
(4) Round Robin Scheduling (cont..)

 Example(7): Consider the following set of processes .

Draw the Gantt chart using (RR) algorithm and find the average waiting
time where time quantum (q=5ms)

Process Arrival time Burst time (ms)

P1 3 18
P2 5 2
P3 9 1

P1 P2 P1 P3 P1 P1
3 8 10 15 16 21 24

 W(P1) =3 ms ; W(P2) = 3 ms ; W(P3) = 6 ms

 Average waiting time = (3 + 3 + 6 )/3 = 4 ms
37
(4) Round Robin Scheduling (cont..)
 If there are n processes in the ready queue and the time
quantum is q, then :
 Each process gets [(1/n) *CPU time] in chunks of at most (q) time units
at once.
 Maximum waiting time for any process= (n-1) *q time units.

 Performance depends on the size of the time quantum (q)

 If q is very large  RR is same as FCFS
 If q is very small  decrease the performance because of context
switch time and increase system overhead

Switching the CPU to another process requires performing a state save of

the current process and a state restore of a different process.

38
(4) Round Robin Scheduling (cont..)

 Turnaround time depends on the size of the time quantum:

 The average turnaround time can be improved if most processes finish

their next CPU burst in a single time quantum.

39
(4) Round Robin Scheduling (cont..)

 RR Scheduling Pros. (++):

 Suitable to time-shared system (better response time)

 RR Scheduling Cons. (--):

 The average waiting time is often long.

 Context switch overhead is higher

40
(5) Multilevel Queue Scheduling

Basic Methodology:
 Processes are classified into different groups (/queues)
 Ready queue is partitioned into separate queues
 Each queue has its own scheduling algorithm.
 Scheduling must be done between the queues (usually priority
scheduling is used)
 The processes are permanently assigned to one queue.
 Example: classify the processes to (2) queues:
o Q1: Foreground (interactive) processes >> scheduled by RR, (high
priority)
o Q2: Background (batch) processes >> scheduled by FCFS, (low priority)
o Scheduling between the queues >> Non-preemptive priority

41
(5) Multilevel Queue Scheduling(cont..)

 Example (8):
(Q1)
 A multilevel queue scheduling algorithm with
(5)queues
 Scheduling between queues is preemptive (Q2)
priority & each queue has a priority over
lower queues.
(Q3)

 No process in (Q4) could run unless the (Q4)

queues for Q1, Q2, and Q3 were all empty.
 If an interactive editing process entered the
(Q5)
ready queue (Q3) while a batch process was
running, the batch process would be
preempted.

42
(5) Multilevel Queue Scheduling(cont..)

 Example(9): Consider a system using (Multilevel Queue Scheduling) algorithm. It

classify the processes to (2) groups (/queues):
o Q1: interactive processes >> scheduled by RR algorithm & (q=4ms)
o Q2: batch processes >> scheduled by FCFS algorithm
o Scheduling between the queues >> Non-preemptive priority
o Q1 has the highest priority.

Draw the Gantt chart for the following set of processes and find the average waiting
time.

Process Process type Arrival time Burst time (ms)

P1 Interactive 0 7
P2 batch 2 5
P3 Interactive 10 2
P4 batch 11 4

43
 Multilevel Queue Scheduling(cont..)
Example(9): Cont.

Q1 P1 P1 busy P3
0 4 7 12 14

Q2 P2 busy P4
7 12 14 18

CPU Gantt chart:

P1 P2 P3 P4

0 7 12 14 18

• W(P1) =0 ms ; W(P2) = 5 ms ; W(P3) = 2 ms ; W(P4) = 3 ms

• Average waiting time = (0 + 5 + 2 +3)/4 = 2.5 ms

44
(5) Multilevel Queue Scheduling (cont..)

 Multilevel Queue Scheduling Pros. (++):

 Consider different process prosperities & requirements

 Multilevel Queue Scheduling Cons. (--):

 Inflexible: a process can’t change it’s queue

 If priority scheduling is used between queues, starvation is
possible and NO solution for it.

45
(6) Multilevel Feedback Queue

Basic Methodology:
 Processes are classified into different groups (/queues)
 Ready queue is partitioned into separate queues
 Each queue has its own scheduling algorithm.
 Scheduling must be done between the queues (usually priority
scheduling is used)
 The processes can be moved between the various queues.

The main
deference to
previous algorithm

46
(6) Multilevel Feedback Queue (cont..)

Example:
 Consider a system separates processes according to the
characteristics of their CPU bursts to (2) groups:
o Q1: I/O-bound processes >> scheduled by RR , (high priority)
o Q2: CPU-bound processes >> scheduled by FCFS, (low priority)
o Scheduling between the queues >> priority scheduling

 A process that waits too long in low-priority queue (Q2) ,the

system will move it to a higher-priority queue (Q1).
 This form of aging prevents starvation.

47
(6) Multilevel Feedback Queue (cont..)
 Example(10): consider a multilevel
feedback queue scheduler with three queues:
 Q0 >> RR , (q=8 ms) ,(high priority)
 Q1 >> RR , (q=16 ms) ,(medium priority)
Q0 ( high priority)
 Q2 >> FCFS ,(low priority)
 Initially, all arrived processes assigned to Q0.
 Processes in lower priority queue is selected if the
higher queues are empty
 A new job enters queue Q0 which is served RR. When it
gains CPU, job receives 8 milliseconds. If it does not Q1 (medium priority)
finish in 8 milliseconds, job is moved to queue Q1.
 If Q0 is empty, process at Q1 job is again served RR and
receives 16 additional milliseconds. If it still does not
complete, it is preempted and moved to queue Q2.

This example approximates the performance of SJF Q2 (low priority)

algorithm without the need to predict the length of

the next CPU bust…(less overhead)
48
Multilevel Feedback Queue (cont..)
 Example(11): Consider the following set of processes
Where the system have (3) queues:
Q1: RR, q=2 ms……….. (highest priority)
Process Arrival Burst
Q2: RR, q=4 ms
time time
Q3: Non preemptive-SJF.. (lowest priority)
P1 3 82 Scheduling between queues is Non-preemptive
P2 4 71 priority
Initially, all arrived processes assigned to Q1.
P3 8 71
In Q1 & Q2 : If the process doesn’t finish in it’s time
P4 12 1 slice, it will be moved to the next lower-priority
queue.
If a process waiting time ≥ 13 ms , it must be
upgraded to the next higher queue
Perform the following:
a) Draw the CPU Gantt chart & find the Average waiting.
b) Find when a process will be demoted.
c) Is there any process will be upgraded ? Which one? When?
49
 Multilevel Feedback Queue (cont..)
Example(11): Cont.

Q1 P1 P2 busy P3 P4
3 5 7 11 13 14

Q2 P1 busy P2 P3 P1 P2
7 11 14 18 22 24 25

Q3 P3
25 26

(a) CPU Gantt chart:

P1 P2 P1 P3 P4 P2 P3 P1 P2 P3

3 5 7 11 13 14 18 22 24 25 26

50
Multilevel Feedback Queue (cont..)
 Example(11): Cont.
TIME W(p1) W(p2) W(p3) W(p4)
(a) Average waiting time =
5 0 1 -- -- (13+14+11+1)/4 =9.75 ms
7 2 1 -- --
11 2 5 3 -- (b) The process will be demoted
13 4 7 3 1 according the following conditions:
14 5 8 4 -- Q1  Q2 : If Process burst time >2
Q2  Q3 : If Process burst time >6
18 9 8 8 --
22 13 12 8 -- (c) P1 has been upgraded at
24 -- 14 10 -- time(22) from Q3 to Q2
25 -- -- 11 --
26 -- -- -- -- P2 has been upgraded at
time(24) from Q3 to Q2

51
(6) Multilevel Feedback Queue (cont..)
 Multilevel Feedback Queue Scheduling Pros. (++):

 Consider different process prosperities & requirements

 Very flexible >>>it is the most general CPU-scheduling
algorithm.
 Can be configured to prevent starvation.

 Multilevel Feedback Queue Scheduling Cons. (--):

 Most complex algorithm

52
Multiple-Processor Scheduling

53
Multiple-Processor Scheduling

 CPU scheduling is more complex when multiple

CPUs are available.

 In multiple-Processor System, load sharing is

possible.

keep the workload balanced among all

processors to fully utilize the benefits of
having more than one processor.

54
Multiple-Processor Scheduling (cont..)
 Approaches to Multiple-Processor Scheduling
Asymmetric multiprocessing Symmetric multiprocessing
(ASMP) (SMP)
Only master processor executes
Each processor can run system code &
system code & slave processors
user code
execute user code
Only the master processor has all
scheduling decisions, I/O processing, Each processor is self-scheduling,
and other system activities
Simple Complex

55
Multiple-Processor Scheduling (cont..)
 Approaches to Multiple-Processor Scheduling (cont..)
Asymmetric multiprocessing Symmetric multiprocessing
(ASMP) (SMP)
only master processor Each processor has its own private queue
accesses the ready queue, reducing of ready processes.
the need for data sharing. OR
All processors take processes from one
common ready queue , So:
• The scheduler must be programmed
carefully.
• It must ensure that two separate
processors do not choose to schedule the
same process.
• It must ensure that processes are not lost
from the queue.

56
Thank you
End of Chapter 4