ch4 - basic

Question 1

Threads run within applications?

Answer 1

True

Question 2

thread creation is (heavyweight)/(lightweight)

Answer 2

lightweight

while process creation is heavyweight

Question 3

draw a single threaded process

Answer 3

+——————————-+
code
+——————————-+
data
+——————————-+
files
+——————————-+
registers
+——————————-+
PC
+——————————-+
stack
+——————————-+
thread → ~~~
+—————————–+

Question 4

draw a multithreaded process

Answer 4

+————————————————-+ code
+————————————————-+
data
+————————————————-+
files
+————————————————-+
registers—-registers—-registers
+————+————-+——————–+
stack ——- stack———stack
+————-+————-+——————–+
PC ————PC ————PC
+————-+————-+——————–+
thread———thread———-thread
+————-+————-+——————–+

Question 5

What is the basic flow of a multithreaded server when handling a client request?

Answer 5

1.Client sends a request to the server.

2.Server creates a new thread to handle the request.

3.Server immediately resumes listening for additional client requests.

Question 6

what are the benefits of Multithreaded Server Architecture?

Answer 6

1.Responsiveness – may allow continued execution if part of process is blocked, especially important for user interfaces
2.Resource Sharing – threads share resources of process, easier than shared memory or message passing
3.Economy – cheaper than process creation, thread switching lower overhead than context switching
4.Scalability – process can take advantage of multicore architectures

Question 7

What is the “dividing activities” challenge in multicore systems?

Answer 7

Breaking a program into independent tasks that can run in parallel without interfering with each other.

Question 8

What is the “balance” challenge in multicore systems?

Answer 8

Ensuring that all processing cores are kept busy and no core is overloaded or idle too long.

Question 9

What is the “data splitting” challenge in multicore systems?

Answer 9

Dividing data into parts so that different threads can work on them independently without conflicts.

Question 10

What is the “data dependency” challenge in multicore systems?

Answer 10

Managing situations where tasks depend on each other’s results, which can cause delays or require careful synchronization.

Question 11

What is the “testing and debugging” challenge in multicore systems?

Answer 11

It’s harder to test and debug parallel programs because bugs may only appear under specific, rare timing conditions (heisenbugs).

Question 12

What does parallelism mean in operating systems?

Answer 12

Parallelism means the system can truly perform more than one task at the exact same time.

Question 13

What does concurrency mean in operating systems?

Answer 13

Concurrency means the system supports making progress on multiple tasks at once, even if not truly executing them simultaneously.

Question 14

Can a single processor provide concurrency? How?

Answer 14

Yes, by using a scheduler that rapidly switches between tasks to give the illusion that they are running at the same time.

Question 15

Can a single processor provide parallelism? How?

Answer 15

NO! it can’t

Question 16

What’s the difference between concurrency and parallelism?

Answer 16

Concurrency = making progress on multiple tasks (may or may not be truly simultaneous).

Parallelism = multiple tasks executing exactly at the same time.

Question 17

Draw concurrent execution on single-core system

Answer 17

|T1|T2|T3|T4|…|T2|

Question 18

Draw parallelism on a multi-core system

Answer 18

core1:|T1|T3|T1|T3|…|T1|

core2:|T2|T4|T2|T4|…|T2|

Question 19

how many types of parallelism?

Answer 19

2

Question 20

What are the types of parallelism?

Answer 20

Data parallelism – distributes subsets of the same data across multiple cores, same operation on each
Task parallelism – distributing threads across cores, each thread performing unique operation

Question 21

What is Data parallelism?

Answer 21

Data parallelism involves splitting a large dataset into smaller chunks, which are processed simultaneously across multiple processors. Each processor works on a part of the data using the same operation, often seen in tasks like vector or matrix computations.

Question 22

What is task parallelism?

Answer 22

Task parallelism divides a program into distinct tasks, where each task performs a different operation on the data. These tasks run concurrently, and each task may have its own specific computation.

Question 23

compare data parallelism and task parallism

Answer 23

Both data parallelism and task parallelism involve executing multiple operations simultaneously to improve performance.

Both techniques aim to optimize computational efficiency by leveraging multiple processors or cores.

Question 24

contrast data parallelism and task parallism

Answer 24

Data Parallelism: Focuses on splitting large datasets into smaller chunks and performing the same operation on each chunk.

Task Parallelism: Divides a program into distinct tasks, each performing different operations, and runs them concurrently.

Data Parallelism: Typically requires less coordination between tasks as they perform the same operation on different data.

Task Parallelism: Often involves more complex communication and synchronization between tasks, as they may depend on one another.

Question 25

what is the purpose of Amdahl’s Law?

Answer 25

It identifies performance gains from adding additional cores to an application that has both serial and parallel components

Question 26

what is Amdahl’s Law equation?

Answer 26

speedup <= 1/(S+(1-S)/N) | S is serial portion N processing cores ## Footnote For example: if application is 75% parallel / 25% serial, moving from 1 to 2 cores results in speedup of 1.6 times

Question 27

To what speedup approaches when N approaches infinity?

Answer 27

1/S

Question 28

# True or False? Serial portion of an application has disproportionate effect on performance gained by adding additional cores

Answer 28

True that's because the serial portion is the part that cannot be parallelized in a program and it must be executed by a single core.

Question 29

Write the names of multithreading models

Answer 29

Many-to-One One-to-One Many-to-Many

Question 30

Explain Many-to-One multithreading model

Answer 30

Many user-level threads mapped to single kernel thread One thread blocking causes all to block Multiple threads may not run in parallel on multicore system because only one may be in kernel at a time

Question 31

Explain One-to-One multithreading model

Answer 31

Each user-level thread maps to kernel thread Creating a user-level thread creates a kernel thread More concurrency than many-to-one Number of threads per process sometimes restricted due to overhead

Question 32

Explain Many-to-Many multithreading model

Answer 32

Allows many user level threads to be mapped to many kernel threads Allows the operating system to create a sufficient number of kernel threads

Question 33

Explain the key difference between the many-to-one and one-to-one multithreading models.

Answer 33

In many-to-one, the OS only sees a single thread, so no true parallelism on multicore systems; in one-to-one, multiple threads can run truly in parallel because each has a separate kernel thread. ## Footnote Many-to-One maps many user-level threads to one kernel thread. One-to-One maps each user thread to its own kernel thread.

Question 34

In the many-to-one threading model, why can a blocking system call in one thread block the entire process? How does this compare to the one-to-one model?

Answer 34

In many-to-one, all user threads are tied to one kernel thread. If one thread makes a blocking system call (e.g., waiting for I/O), the single kernel thread is blocked, causing all user threads to stop until the system call finishes. In one-to-one, each user thread has its own kernel thread — so if one thread blocks, other threads can still continue execution independently.

Question 35

Describe how the many-to-many multithreading model improves upon the limitations of both many-to-one and one-to-one models

Answer 35

Many-to-Many maps many user threads to a smaller or equal number of kernel threads, flexibly. It avoids the bottleneck of many-to-one (where a blocking call stops everything) and avoids the heavy overhead of one-to-one (where each thread demands a kernel thread). Advantage: Lets multiple threads execute concurrently without flooding the system with too many kernel threads.

Question 36

What is Implicit Threading?

Answer 36

Creation and management of threads done by compilers and run-time libraries rather than programmers

Question 37

Explain Thread Pools

Answer 37

Create a number of threads in a pool where they await work **Advantages: **Usually slightly faster to service a request with an existing thread than create a new thread Allows the number of threads in the application(s) to be bound to the size of the pool Separating task to be performed from mechanics of creating task allows different strategies for running task

Question 38

Explain Fork Join

Answer 38

divide to subtasks, then build the result from the results of the tasks | Same idea as divide and conquer

Question 39

Mention the names of threading issues

Answer 39

Semantics of fork() and exec() system calls Signal handling (Synchronous and asynchronous) Thread cancellation of target thread (Asynchronous or deferred) Thread-local storage Scheduler Activations

Question 40

Why signal handling is a threading issue?

Answer 40

n multithreaded programs, who handles the signal? All threads? One thread? Which one? You have to manage signals carefully across threads...

Question 41

Why Thread cancellation is a threading issue?

Answer 41

If you cancel a thread, it could be holding locks, halfway through an operation, writing to a file... Canceling threads can cause resource leaks, deadlocks, and data corruption if not handled properly.

Question 42

Why Thread-local storage is a threading issue?

Answer 42

Thread-local storage lets each thread have its own private copy of data If you don’t use it properly, threads might corrupt shared data or accidentally stomp on each other’s stuff.

Question 43

Why is Scheduler Activations a threading issue?

Answer 43

How do we schedule threads when the kernel doesn’t even know all the user threads exist?