CSD 204 Lab
Lab 6: Implementing TAS, CAS and Bounded
Waiting CAS Mutual Exclusion Algorithms
[Dr. Sweta Kumari, Assistant Professor, SNIoE]
Deadline – 27th Mar 2025, 11:59PM
Goal:- The goal of this assignment is to implement TAS, CAS and Bounded Waiting
with CAS mutual exclusion (ME) algorithms studied in the class. Implement these
algorithms in C++.
Details:- As shown in the book, you have to implement the three mutual exclusion
algorithms in C++. Each mutual exclusion algorithm has an entry and exit sections.
To test the performance of mutual exclusion algorithms, develop an application, me-
test (mutual exclusion test) is as follows. Once, the program starts, it creates n
threads. Each of these threads, will enter critical section (CS) k times. The
pseudocode of the test function is as follows:
Here t1 and t2 are delay values that are exponentially distributed with an average of
λ1, λ2 milliseconds. The objective of having these time delays is to simulate that
these threads are performing some complicated time-consuming tasks.
Input: The input to the program will be a file, named [Link], consisting of
all the parameters described above: n, k, λ1, λ2. A sample input file is: 100 100 5
20.
Output: Your program should output to a file in the format given in the pseudocode
for each algorithm. A sample output is as follows:
TAS ME Output:
1st CS Requested at 10:00 by thread 1
1st CS Entered at 10:05 by thread 1
1st CS Exited at 10:06 by thread 1
1st CS Requested at 10:01 by thread 2
CAS ME Output:
1st CS Requested at 11:00 by thread 1
1st CS Entered at 11:05 by thread 1
1st CS Exited at 11:06 by thread 1
1st CS Requested at 11:01 by thread 2
Bounded CAS ME Output:
1st CS Requested at 11:30 by thread 1
1st CS Entered at 11:35 by thread 1
1st CS Exited at 11:36 by thread 1
1st CS Requested at 11:32 by thread 2
.
The output should demonstrate that mutual exclusion is satisfied. The output could
be in a single file or multiple files.
Report: You have to submit a report for this assignment. This report should contain
a comparison of the performance of TAS, CAS and Bounded CAS ME algorithms.
You must run these algorithms multiple times to compare the performances and
display the result in form of a graph.
For performance, you have to specifically measure two metrics: (1) the average time
taken by a process to enter the CS (2) the worst case time taken by a process to enter
the CS in a simulation. This shows if processes are starving.
You run these algorithms varying the number of threads from 10 to 50 while keeping
other parameters same. Please have k, the number of CS requests by each thread,
fixed to 10 in all these experiments. The graph in the report will be as follows: the
x-axis will vary the number of threads. The y-axis will show the metrics defined
above: (1) average time taken to enter the CS by each thread (2) the worst-case time
taken by a process to enter the CS in a simulation.
Also ensure that each point of the graph is obtained by averaging over five runs.
Finally, you must also give an analysis of the results while explaining any anomalies
observed.
Submission Format:- You have to upload: (1) The source code in the following
format: Assgn6Src-<Name>.c (2) Report: Assgn6Report-<Name>.pdf. Don’t zip
any of your documents. Upload all the documents on the blackboard.
Note: Please follow this naming convention mentioned above.
Grading Policy:- The policy for grading this assignment will be - (1) Design as
described in the report and analysis of the results: 50% (2) Execution of the tasks
based on the description in the readme: 40% (3) Code documentation and
indentation: 10%.
Please note:
- All assignments for this course have a late submission policy of a penalty of 10%
each day after the deadline of six days. After that, it will not be evaluated.
- All submissions are subject to plagiarism checks. Any case of plagiarism will
be dealt with severely.
Sample Questions:
1. Write a program using C and pthread library to perform a parallel matrix
multiplication routine. The goal is to multiply an MxN matrix (A) by an NxP matrix
(B) and then store the result into the MxP matrix (C). You can design your program
in such a way that each thread does an equal share of the work or you can have the
threads run in a loop that computes single rows of the result (i.e. C). Also print time
taken by each thread and compare the time taken by C program with threads (parallel
program) and without threads (sequential program). Use gettimeofday() to measure
the time.
Test Cases-
Input
int A[3][3] = {{1, 2, 3}, {4, 5, 6}, {7, 8, 9}};
int B[3][3] = {{9, 8, 7}, {6, 5, 4}, {3, 2, 1}};
Output
Resultant Matrix:
30 24 18
84 69 54
138 114 90
Input
int A[3][3] = {{1, 0, 0}, {0, 1, 0}, {0, 0, 1}};
int B[3][3] = {{5, 6, 7}, {8, 9, 10}, {11, 12, 13}};
Output
Resultant Matrix:
5 6 7
8 9 10
11 12 13
2. Calculate the addition of all the elements in a single array using multiple threads.
Use mutex to implement a critical section so that no two threads simultaneously
update the global result variable (that stores the final addition result).
Input
{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}
Output
Parallel execution sum: 55
Sequential execution sum: 55
Q.3. Implement a parallel search algorithm where multiple threads search for an
element in a large array or list.
Q.4. Implement a parallel merge sort algorithm using threads. Divide the array into
smaller parts and sort them concurrently.
Q.5. Implement a counter that is incremented by multiple threads using atomic
operations instead of mutexes.
