Operating System Concepts and

Techniques Lecture 2

Multiprogramming, Multithreading, Multiprocessing, and Multitasking

M. Naghibzadeh

ReferenceM. Naghibzadeh, Operating System Concepts and Techniques, First ed., iUniverse Inc., 2011.

To order: www.iUniverse.com, www.barnesandnoble.com, or www.amazon.com

2

DefinitionsProgram: a set of instructions

which is prepared to perform a specific assignment if executed by

a computer. Program need not be connected; it could be stored on a flash memory

and placed in one’s pocket.Program is not an active entity. It is

completely passive.

3

Definitions…Process: is created by operating system

from a program To do so, it has to perform many activities,

like assigning a name, allocating space, (partially) loading the corresponding

program, etc.

Roughly speaking, A process is an active program

A process is created to run a program by using computer facilities such as CPU;

like a human who is born to live his life

4

Single programmingSingle-programming: there exist at the most one user process at any given time in the computer;

thus there is usually one ongoing activity at a time

That is, from many devices within the computer often one device is active at any given time

If a process has asked for a data from the user, the system has to wait until this data is entered before being

able to proceed If this data entry takes one second the CPU could have

done millions of instructions if it did not have to wait

With single-programming the computer resources are not used efficiently

5

Process States in Single-Programming

With single-programming, right after a process is born the system starts running it

It continues until the process has to wait for data, output results, or wait for an event

There are special purpose processors called Input/Output (I/O) processors for transferring data from input devices to

main memory and from main memory to output devices It is understandable that such a processor will perform the

specific task better than a general-purpose processor, i.e., CPU

While an I/O operation is in progress, the CPU has to wait and do nothing. After the I/O operation is completed, the

CPU will resume the execution of the programThis cycle, of going through process states of running and input/output, may be repeated over and over, until the job

is completed or, for some reason, the process is abortedThe life cycle of a process in a single-programming

environment is shown in Figure 1

6

Process’s life cycle

Figure 1: The life cycle of a process in single-programming environments

Process birth Running Input/Output

Process Termination

Blocked for I/O

I/O completed

7

Processor wait ratio

be

bw

If the average execution time of a program with single-programming is e and the

average I/O time is b, then the following ratio is the CPU wait fraction (w). It is

actually the fraction of the time the CPU is idle.

For example, if execution time of programs is 10, of which 9 seconds is spent on I/O, then w = 9/10 = 0.9. This means, on the

average, 90% of the CPU time is wasted.

8

The Multiprogramming Concept

Multiprogramming: is a technique that allows more than one program to be ready for execution

(process) and provides the ability to switch from one process to another, even if the former is not

completed Of course, sometimes in the future we will have to

switch back to the first process and resume (not restart) its computation

This technique works for both single-processor (like some personal computers) and

multiprocessor (such as large main frame) computers.

Multiprogramming is mainly accomplished by the operating system. The hardware provides some

specific circuitry that may be used by the operating system in the course of facilitating

multiprogramming

9

Multiprogramming and PCs

Do we need multiprogramming for PCs? Yes. All PC users like to run many applications

simultaneously Nobody runs for example an Internet explorer looking

for an information while staring at the monitor for the results for a long time.

In this era of computer usage, every general-purpose operating system must have the

following capabilities: It must provide an environment to run processes in a

multiprogramming fashion. It must act as a service provider for all common services

that are usually needed by computer users, such as copying files, making new folders, compressing

information, sending and receiving messages from/to other computers in the network, etc.

Its user interface must be easy to use and pleasant to work with.

10

Multiprogramming productivity

Multiprogramming increases productivity If CPU wait time is represented by w in single-

programming environment, the CPU wait time decreases to approximately wn for a system

running n processes simultaneously. Example: If w = .9 then wn =0.59 ; meaning

that if we have five processes running simultaneously, the CPU utilization is

increased by (0.41-.10)/0.1*100 = 310%.

By increasing the CPU utilization other device’s utilization is also increased.

11

Process State Transition Diagram

The life cycle of a process in multi-programming has three states

Ready: ready to use CPU, however CPU is busy

Running: Uses CPU Wait/Blocked

Wait: process is waiting for a device or an event Blocked: process is waiting for its I/O to be

completed by an I/O processor

May switch between states many times during its life

12

Process’s life cycle

Figure 2: Basic process state transition diagram in multiprogramming

Process birth

ReadyWait/Blocked

Process Termination

Running

A process is picked to run

Needs I/O or circumstance

Running obstacle is vanished

Preempted for the interest of others

13

Requirements of Multiprogramming

Process Switching possibility: the system must be able to safely switch from one process to

another and be able to return in the future This is called context switching.

Direct Memory Access: I/O processors must be able to directly access main memory without

interference and conflictions The Interrupt System: I/O processors and

monitoring devices must be able to safely communicate with the CPU to report or to get

new assignments

To context switchSave enough information from current

process to be able to return to it and continue

Save all temporary storage information such as registers and flags

Load/initialize the information of new process

Change PC Need some help from Hardware to make

the switching process fast

14

15

InterruptInterrupt: a signal that is sent to the CPU to capture its attention. It may be issued

by different sources such as: By a fault detecting hardware that is trying to

informs the CPU of a circuitry malfunction By a monitoring circuit within the ALU that controls the acceptability of the data size for

the operation being performed and, consequently, the operation result size

By a user that may press a special key like “break” at any time

By an slave processor that wants to inform the CPU of the completion of the assigned

task By a timer etc.

16

Interrupt masking Some interrupts are maskable by user

programs Masked interrupts are ignored by the CPU

This is analogous to ignoring a door knock

Some interrupts are not user-maskable It is not reasonable to let the computer

continue in the presence of a hardware failure.

However, all interrupts are maskable within the kernel of an operating system.

17

Interrupt CategorizationInterrupts are categorized into few

classesDifferent classes of interrupts require

different handling considerationEach category of interrupts is represented by a bit in interrupt vector within the CPU

Each interrupt category has a priorityWhen handling a higher priority interrupt

handling any lower priority interrupt is postponed

18

Program execution suspension

A non-masked interrupt causes the suspension of the currently executing program.

Usually possible to resume the execution of the program after the interrupt is handled.

Interrupt signals reach the CPU and are stored in the interrupt vector. The interrupt(s) is ignored until the

currently executing machine instruction is completed. There are very few exceptions. One example might

be the “move” instruction in some computers that is suppose to move a big chunk, say 256 bytes, of data

from one location of main memory to another. The CPU always looks for interrupt signals just before

fetching a new instruction to execute. If there are any interrupt signals, the system handles them in the

order of priorities and urgencies.

19

Fetch-Execute Cycle with interrupt

Figure 4: A simplified functional model of a computer with interrupt

Read the instruction pointed out by the Program Counter (PC) register from main memory and move it to the CPU

Find out what this instruction is

Adjust PC for the next instruction

Fetch cycle

Execute the instruction, i.e., perform what is requested by the instruction. Perhaps this may readjust the PC.

Execute cycle

Move the data upon which the instruction has to be executed from main memory to the CPU.

Handle all non-masked interrupts

20

Privileged instructionsPrivileged instructions are those which are

not usable by assembly application programmers

Privileged instructions are executable only within the operating system kernel

Other parts of the OS and application programmers may execute privileged instruction by special means that are

provided by the operating systemDisable interrupt which disables all

interrupts is a sample privileged instruction

21

MultiprocessingIf you think clearly, you will notice that we

should have used multiprocessing instead of multiprogramming. This is true.

Unfortunately, the term “multiprogramming” is recognized for this technique of the operating system and we

will stick to itOn the other hand, “multiprocessing” is

used for systems with more than one processor. Processors, in such a system, can

collectively run many tasks simultaneously

22

MultitaskingComputer users like to have many application programs

simultaneously operational. This is necessary because some application programs require long processing times before the

desired results can be produced. It is true that by having more than one application program

operational, the time that it takes for each process to complete its task increases. However, the overall system productivityThese simultaneously executing programs are called tasks.

Therefore, a system with the capability of multitasking allows users to activate more than one task, or application program,

at a time. An Internet browser that searches for some information and A word-processing software that is activated to

perform the word-processing task are applications. The operating system will switch between tasks based on the

tasks current states and their requirements and priorities. Multitasking is only possible when multiprogramming is the fundamental capability of simultaneously executing pieces of

software.Most modern operating systems, like UNIX, Linux, and

Windows, support multitasking.

23

MultithreadingA multithreading operating system is one that is capable of handling processes and threads at the same time and

in which from each process the system is able to generate more than one thread.

In such an operating system, there must be facilities for thread creation, deletion, switching, etc.

Such an operating system allows users to generate more than one request to a process at a time. For example, a

browser can be made to search simultaneously for more than one topic, even though there is only one copy of

the “browser program” in main memory.The multiprogramming methodology and technique are

essential in the implementation of multithreading. In this new environment, a thread becomes the smallest

functional object to which CPU (or a PU) is assigned. Details of thread methodology and technique is

discussed in upcoming lectures.

24

SummaryThe ultimate goal in the design and implementation of an

operating system is to produce a handy software program that manages computer resources in the most efficient way so as to

serve computer users correctly, reliably and fairly.This is not achievable in single-programming environments.

Modern operating systems are built with the capabilities of multiprogramming, multitasking, and multithreading.

Providing these capabilities requires many hardware and software methodologies and techniques.

A good understanding of process creation, life cycle, and termination, along with its state transition conditions is most

essential in elucidating the needs of different mechanisms within the operating system.

Some of these mechanisms, namely process switching, interrupt system and handling, and direct memory access, are

briefly explained in this chapter.Multithreading, as an offspring of multiprogramming, has

become an essential part of all modern operating systems.

25

Find outHow many interrupt classes your

computer’s processor hasSome privileged instructions and the

reasons for being privilegedHow are interrupts actually masked

The exact differences of multiprogramming and multiprocessing

What an ultra DMA isThe exact differences between program

and process

26

Any questions?