|
Module code and Title
|
DTS202TC Foundation of Parallel Computing
|
|
School Title
|
School of AI and Advanced Computing
|
|
Assignment Title
|
Individual Assessment 2
|
|
Submission Deadline
|
Friday Dec. 20th, 2024 @ 11:59pm
|
|
Final Word Count
|
N/A
|
DTS202TC Foundation of Parallel Computing
Individual Coursework 2
Assessment Overview
The purpose of this assignment is to provide hands-on experience with parallel programming and performance analysis. You are expected to implement two C programs that apply a blur effect to a grayscale image using two distinct parallelism techniques. The goal is to understand the differences between the techniques in terms of implementation complexity, execution speed, and scalability. Through this exercise, you will develop skills in writing parallel code, optimizing algorithms for parallel execution, and analyzing the performance gains achieved by parallelizing computational tasks.
Learning Outcomes
. C. Devise and implement parallel algorithms
. D. Acquire basic software development skill using MPI
. E. Analyze and implement common parallel algorithm patterns in a parallel programming model such as CUDA.
. F. Design experiments to analyze the performance bottlenecks in their parallel code.
. G. Apply common parallel techniques to improve performance given hardware constraints.
. H. Use a parallel debugger to identify and repair code defects; Use a parallel profiler to identify performance bottlenecks in their code.
. J. Demonstrate understanding of the major types of hardware limitations that limit parallel program performance.
Avoid Plagiarism
. Do not submit work from others.
. Do not share code or work to other students.
. Do not read code or work from others, discussions between teams should be limited to high level only.
. Do not use open-source code.
Tasks
Image blurring is one of the most common image processing algorithms (right plot). A simplest way to blur an image is to set each pixel to the mean of its neighbour pixels. One of the problems of this algorithm is that the effect may not be visible for large images, to solve this, we can repeat the process multiple times. For our coursework, please repeat the mean process 20 times.
Reading and Writing Image
You will need to read a PGM image, process the blurring, then write the blurred PGM image back to the file system. Plain PGM format is a simple grayscale graphic image format, each pixel is represented by its grey value number, with 0 being black and Maxval (defined in the PGM file) being white. The below image.pgm is given as an example, more detail pgm specifications can be found at
http://davis.lbl.gov/Manuals/NETPBM/doc/pgm.html
image.pgm
P2
24 7
15
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 3 3 3 3 0 0 7 7 7 7 0 0 11 11 11 11 0 0 15 15 15 15 0
0 3 0 0 0 0 0 7 0 0 0 0 0 11 0 0 0 0 0 15 0 0 15 0
0 3 3 3 0 0 0 7 7 7 0 0 0 11 11 11 0 0 0 15 15 15 15 0
0 3 0 0 0 0 0 7 0 0 0 0 0 11 0 0 0 0 0 15 0 0 0 0
0 3 0 0 0 0 0 7 7 7 7 0 0 11 11 11 11 0 0 15 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 Parallel implementations (25 points for each implementation)
Each student is required to implement two different parallel programs: one using MPI and the other using a different parallel technique such as Pthreads, OpenMP, or CUDA. It is important to note that you only need to focus on optimizing the core algorithm for speedup.
2 Performance analysis (30 points)
Analyse the performances of two parallel implementations, including speedup and efficiencies (15 points).
You should provide line plots to represent the results, including serial runtime, runtime of different processes, speedup and efficiencies. (5 points).
Compare and contrast the two implementations (Performance Metrics, Hardware limitation, Ease of Debugging and Testing, Communication and Synchronization, or any other aspects you can think of), choose your preferred parallel technique and justify your choice (10 points).
3 Reflection (10 points)
Did you face any challenges during the implementation? How did you solve them?
4 Submission (10 points)
You should submit the following files in a zip:
. blur_mpi/pthread/openmp.c Your parallel implementations.
. A Makefile that will compile your code, make sure the output executable names are correct.
. A pdf file contains all the source code and the report, you should put the Cover Page in the first page of the report.
Once you have all the files, please put them in a single directory (named A2) and compress it to a zip file. You must follow the following structure:
The assignment must be submitted via Learning Mall to the correct drop box. Only electronic submission is accepted and no hard copy submission. All students must download their file and check that it is viewable after submission. Documents may become corrupted during the uploading process (e.g. due to slow internet connections). However, students themselves are responsible for submitting a functional and correct file for assessments.
Please note that quality of report and correctness of submission will also be marked (10 points, 5 points for the quality of report, 5 points for the correctness of submission).