OS topic 19 – The operating system zoo

Question 1

There are a few of the main types of OS that you may be less aware of:

M________________
Em____________ OSs
Real-t________ OSs
P____________ computing systems

Answer 1

Microkernels
Embedded OSs
Real-time OSs
Parallel computing systems

Question 2

Microkernels

Most desktop operating systems use a m________ kernel. This architecture places the entire operating system within the kernel space. This has s_______ advantages, but means that kernels can become huge – millions of lines of code – making them hard to m________. Additionally, a fault anywhere in the kernel can spread to other modules, causing the entire system to f_____.

Answer 2

monolithic
speed
maintain
fail

Question 3

Microkernels

An alternative to the monolithic kernel is the microkernel.
The core principle behind a microkernel is that only a m________ amount of s_______ is provided within the k______, with other operating system functions – such as device drivers and files systems – running in u______ space.

Answer 3

minimal
software
kernel
user space

Question 4

Microkernels

The advantages of a microkernel come at a cost.
What is the cost?

Answer 4

Performance

Question 5

Microkernels - Performance issues

Having to constantly send and receive inter-process communication messages through the microkernel, requiring a switch from user to kernel mode, is an o________ that can cause p___________ issues.

Answer 5

overhead
performance issues

Question 6

Definition - Overhead

In computing, overhead refers to the
e______ or a_______ resources, t_____, or
c____________ power required by a process, system, or operation beyond what is strictly necessary to perform the core functionality.

Answer 6

excess or additional resources
time
computational power

Question 7

Embedded and real-time OSs

Embedded systems are computers that have a d_________ function, sometimes within a larger system.

Give some examples of devices with embedded systems.

Answer 7

dedicated function

smartwatches and mp3 players
traffic-light controllers
cars and medical equipment , etc

Question 8

Embedded and real-time OSs

True or false?
Unlike a general-purpose operating system, the OS of an embedded device has one purpose: to support the running of the software that allows the embedded device to fulfil its purpose.

Answer 8

True

Question 9

Embedded and real-time OSs

There are two key criteria for any embedded operating system:

R________ efficiency and r__________.

Answer 9

Resource efficiency and reliability

Question 10

Embedded and real-time OSs

True or false?
Embedded systems do not have many resource constraints.

Answer 10

False - In addition to a requirement to be power efficient as many are battery powered, embedded systems tend to have small amounts of memory and low-performance CPUs. This helps keep the costs of devices low, and also reflects that the system only needs to undertake certain activities. These resource restraints mean that an embedded operating system has to be as efficient as possible and run on a small memory footprint.

Question 11

Embedded and real-time OSs

Updating embedded systems is not easy.
Why not?

Answer 11

Not all devices have internet access

Question 12

Embedded and real-time OSs

As a result of update constraints, it is essential that the operating system can run r_______ for long periods of time – up to y_______ in some cases – without developing any f_______.

Answer 12

reliably
years
faults

Question 13

Embedded and real-time OSs

Linux or Unix?

_________ is commonly used in many types of embedded systems, including all Android phones and tablets, a wide variety of cars, and a host of other commercial and industrial embedded systems.

Answer 13

Linux

Question 14

Real-time OSs

A real-time system (RTS) has to produce the c______ c_____________ r__________ within a defined time period.

Answer 14

correct computational result

Question 15

Real-time OSs

Real-time systems are often talked about in terms of ‘hardness’ where:

A hard RTS guarantees that critical tasks are completed before the deadline.
If the task is not completed, the system f_____.

A soft RTS gives critical tasks p______ until they are completed. The system doesn’t fail if the task isn’t completed on time.

Answer 15

fails
priority

Question 16

Real-time OSs

Give an example of where a soft RTS might be used.

Answer 16

In systems such as media streaming: it is not particularly noticeable if an audio stream loses a few bits.

Question 17

Real-time OSs

Give an example of where a hard RTS might be used.

Answer 17

A system within an autonomous car deciding whether to brake to avoid the pedestrian who has stepped into the road must deliver a result within a defined time frame, otherwise the result of the decision is irrelevant.

Question 18

Real-time OSs

True or false?
Linux was not developed to be a real-time operating system.

Answer 18

True

Question 19

How did Linux become a real-time operating system?

Answer 19

A patch to the kernel, the PREEMPT_RT patch, has been developed to turn Linux into a real-time operating system.

Question 20

Parallel computing

Since the early 2000s, most processors have become m_______, essentially placing multiple processors onto a single chip so that they can operate in p_______, each completing a separate task. But even this shift in chip design has not kept pace with the processing power needed for some activities.

Answer 20

multicore
parallel

Question 21

Parallel computing

A system with more than one processor should have a higher t___________ of instructions than a single-processor system, since each processor can be executing instructions at the same time. This is true parallel processing – that is, simultaneous performance of r________ tasks by two or more independent and non-dedicated processors – rather than the concurrency obtained by multitasking

Answer 21

throughput
simultaneous
related tasks

Question 22

Parallel computing

True or false?
Adding processors to a system is a simple matter.

Answer 22

False

Adding processors to a system is not a simple matter: there are trade-offs involved which can be satisfied in several different ways.

Question 23

Multiprocessor and multicore systems

What is the difference between Multiprocessor and multicore systems?

Answer 23

The key distinction lies in whether the multiple processing units are integrated onto a single chip (multicore processor) or if they are separate, individual processor chips within the same system.

Question 24

Processors in GPUs

GPU stands for G__________ P__________ U___________.

Answer 24

graphics processing unit

Question 25

Processors in GPUs

The CPU isn’t necessarily the only component of a computer that can p_____ d____.
Another common component able to handle processing is the graphics processing unit (GPU).

Answer 25

process data

Question 26

Processors in GPUs

True or false?
Rendering graphics, particularly 3D graphics, is computationally expensive in terms of processing data.

Answer 26

True

Question 27

Processors in GPUs

GPUs are designed to be extremely m______________ and to process large volumes of data quickly, but are poorly suited to general-purpose processing.

Answer 27

multithreaded

Question 28

Cluster systems

A common a_____________ for increasing processing power is a cluster.

A cluster architecture connects together individual computers – minimally consisting of a processor, local memory and storage – over a fast connection.

A layer of m_________ software sits on each individual computer (also called a n_______), allowing a user to treat the cluster as a single computer.

Answer 28

architecture
middleware
node

Question 29

Cluster systems

True or false?
Usually, each of the computers in a cluster runs the same operating system.

Answer 29

True

Question 30

Cluster systems

True or false?
In a cluster, processors share memory

Answer 30

False - they don’t share memory so the operating system and application programs have to be designed to use other mechanisms for communication; these are significantly slower than in a shared-memory multiprocessor system.

Question 31

Cluster systems

The most sophisticated clusters are s________________ using specialised connections such as the InfiniBand networking connection and running specially designed operating systems.

Answer 31

supercomputers

Question 32

Cluster systems

Supercomputers are designed to run large-scale a_____________ such as climate modelling, drug discovery and weather f_______________

Answer 32

applications
forecasting

Question 33

Distributed systems

Building a supercomputer from large numbers of processors is certainly possible for those who have the money and space to do so. But people have noticed that large numbers of connected processors already exist in the form of n__________ m_________, particularly if they think big and include all the computers connected to the internet.

Answer 33

networked machines

Question 34

Distributed systems

A distributed system is formed when many computers cooperate on a problem, communicating by means of standard n__________ h__________.

Answer 34

network hardware

Question 35

Distributed systems

One of the most famous large distributed systems was the SETI@home project.

This research project based at the University of California at Berkeley is a search for extra-terrestrial intelligence

Volunteers download a program to their machine, and this program runs whenever their machine is otherwise i______.

The SETI program downloads a chunk of data from Berkeley, analyses it, and sends back the results.

Answer 35

idle

Question 36

Distributed systems

True or false?

These projects basically use your computer as a processor.

Answer 36

True

Question 37

Distributed systems

As with a cluster system, there is no shared memory in a distributed system. Instead, communication between processes is by messages sent over conventional networking hardware.

What is a downside to the message system?

Answer 37

Slow