Problem solving essay on Operating Systems paper- Computer science
Matrix Multiplication
Working Summary: In this assignment, you will do:
(a) A multi-threaded matrix multiplication.
(b) and, you will compare the execution times for different numbers of threads
(starting from 1 to 100, incremented by 10).
Name of the program: The name of the program will be ‘MatMul_thread’.
Generation of input matrices: You will create two randomly generated 100 by 100
(square) matrices of real numbers, numbers can be both negative and positive.
Measuring execution time: You should measure the execution time by recording the
start and end time. You must take average execution time from (minimum) 25
iterations for each individual case for number of threads starting from 1 to 100,
incremented by 10. Save your computed results in comma separated vector (CSV)
file named ‘exe_time.csv’, which will contain the average execution time for each
case. You can open such ~.csv file to be processed by Excel for graph generation
easily. Anyway, you will generate a graph: ‘number of threads versus execution
time’, where x-axis indicates number of threads = 1, 10, 20, …, 100 and y-axis
indicates the corresponding average processing time.
Report: Place the above graph in your report document named as ‘myreport’
(preferably an MS Word file, e.g., doc/docx). Explain the graph. If there is any
anomaly, explain that as well. Further, describe your algorithm in pseudo code and
place it in ‘myreport’. Mention your load balancing strategy–how you distributed
data/task among threads. Also, describe the machine you used for running your
program, in terms of number of available processors or cores. Save all input and
output (resultant) matrices in input1.txt, input2.txt, and output.txt files.
Programming Language Preferences: You may use the POSIX, NOT OpenMP for
creating threads. You must code in Java or C/C++.
For any other language preferences consult the instructor (me) first or, make sure
that the language supports the thread or process creation. And if you need, you can
also use the programming language’s libraries related to this parallel computation
(if exists any).
Submission: Compress all the related files in a single folder and submit. You must
submit the source code, executable version including all the files mentioned before –
the code must be well documented/commented. You must include a ‘readme’ file
which will describe how to run your program, plus specify the compiler used for the
code. Describe the POSIX functions in the ‘myreport’ that you have used in your
program.
Due: 1 May 2018 11:55pm
Late submission cutoff: 5 May 2018
(Work submitted after regular deadline but before the late submission cutoff will be
graded at 85% max.)
