The purpose of this project is to familiarize yourselves and your classmates with recent and important research topics and advances in the field of parallel computation. Your project is intended to be a research-oriented study or an actual implementation. It has several intentions for purposes of your education in this class, among them including improvement of your ability to prepare for further graduate or undergraduate-level research, improving familiarity with recent advances, self-study in a a particular area of interest, and the practice at your written communication skills.
Projects are intended to be completed by groups of two or three students, but in certain cases, single-person projects are also acceptable. You will be graded in a number of different areas, but the primary criteria include your overall body of work, the organization of your work, and the clarity of your written presentations.
You should initially decide upon a topic on which you will focus your research project. It will either be a general area to investigate or an implementation on a parallel computer. Please sign up with me by giving me an idea of your intended topic area.
If you sign up in the same topic area as other students, you may form a team that collaborates on the project. For a team project, it is expected that you will be able to produce a more elaborate or productive project, but it will also have the benefit of reinforcing each other's work in your area of choice.
The list (in the box below) is a set of possible suggestions and is not meant to be comprehensive. Therefore, you are encouraged to also think outside of the box. If you would like to examine other topics or specific sub-areas, then please contact me to discuss.
Possible Topics
Parallel Architectures
Parallel Languages
Parallel Compilers
Parallel Algorithms
Implementations of Parallel Algorithms
Parallel Simulation
Performance Modeling
Synchronization Mechanisms
Operating Systems for Parallel Processing
Different Computational Platforms (tightly coupled vs. grid, peer-to-peer, cluster, etc.)
Cloud computation
Distributed Parallel Algorithms
Specific programming projects for specific architectures.
Topics you may have thought of in homework 1, problem 1
Project Approach
You can choose to do an implementation or a research project. Each of these requires a report, although it's expected that a research project will involve a more in-depth study (it will make use of more references and the report will be considerably greater in depth, and probably, length).
Research Report:
Propose a topic to study in depth from the list in the table above or from some other approved topic.
If you choose this option, then as an approximate guideline, graduate students are expected to turn-in 20-30 pages, while undergraduates are expected to turn-in 15-20 pages. It is expected that your work reflects having done extensive research on what is considered to be the best practices or state-of-the-art for your specific subject. You should try to find new approaches published recently (within the last 2-3 years if possible).
Project Report:
Propose a project to be implemented on one or more parallel computers on which you have access. You may choose to implement an algorithm using different languages on the same or different machines, measure performance of different implementations, develop a set of bench marks, study system effects of specific architectures on performance such as caching, paging, latency, etc.
You still have to write a paper that summarizes your project, but it can be shorter than that of the research paper variety. Other than the design aspects (which could be a larger amount of text/images), I expect your write-up to be between 5 and 15 pages.
Programming projects must also include copies of the source code, test data sets, as well as instructions for running the program(s). For these reports, you must explain your methods, findings, any comparisons, and your learning experience. Include performance measurements.
Required minimum 10 page