Computer Science 400
Parallel Processing and High Performance Computing

Fall 2017, Siena College

Home     Syllabus     Schedule

You may work individually, but are encouraged to work in groups of two or three. A group of size four could be approved for an especially large project. Groups must be formed by Tuesday, October 31, 2017.

Since everyone will be working on a different project, it would be appropriate to make your GitHub repositories public, to help build your code portfolio. You may choose to include your written materials in the public repository or keep those separate.

The proposal

Please start thinking about a topic right away and come to me with your ideas. By 3:30 PM, Tuesday, November 7, 2017(but hopefully much sooner), submit a proposal, at most one page in length, that describes your topic, what specifically you plan to investigate about your topic, and how you plan to go about it. Describe the major milestones for your project, a rough schedule for achieving these milestones, and which milestones you believe are most important for your project to be considered a success. If you will need access to any special hardware or software, include that in your proposal. Your proposal should convince me that you have an interesting, worthwhile, and relevant topic and that it is feasible in the time available.

The progress report

By 3:30 PM, Tuesday, November 21, 2017, submit a progress report. This should consist of a detailed outline (or better yet, an early draft) of your paper including sources, and a description of the design and current implementation status of your software, as well as a more specific timetable for completion of the project. You must also submit an annotated bibliography with your progress report, detailing the sources you have considered (including some you intend to use and some you do not intend to use).

The paper

This is to be a formal research paper, and should be organized as such. You should begin with a title, author list, and abstract. The main body of the paper should be organized into sections including (i) an introduction in which you describe the general topic and the particular aspects you will be examining, (ii) one or more sections comprising your main text, where you describe what you have done, how you have done it, and what you have learned, (iii) a conclusion, which should include ideas for future investigation into your topic which were beyond the scope of your project and the paper, and (iv) a complete list of citations. Citations of web pages are acceptable in some circumstances, but books, articles in conference proceedings or journals, or technical reports are preferred.

Proper English and a good technical writing style are important. Writing well is very difficult - it is an iterative process and cannot be done all at once. Be precise and be concise. Group members should proofread and make suggestions about each other's writing. Check your spelling and grammar carefully. I expect most papers will be around 15 single-sided pages, using 1.5 spacing, one inch margins, and a 12-point Times Roman font (or similar). Please do not adjust margins and font sizes to force a certain length. You are encouraged but not required to use LaTeX to typeset your paper. The LaTeX example in a public GitHub respository is already set up with an appropriate format. Length is not important - content and quality are. Papers shorter than 10 pages or longer than 20 pages are acceptable, if the length is appropriate for the content.

Submit a complete draft by 3:30 PM, Monday, December 4, 2017. The more complete this draft, the better feedback I can give you to improve it and the more you can focus on wrapping up any programming tasks and preparing your presentation at the end. You may also submit additional drafts for feedback, but keep in mind that it may take me a day or two to get to them. Submit your final version by 5:00 PM, Monday, December 11, 2017.

The project

You should submit your source code and instructions on how to build and run it. The software should be described in the paper. Please make arrangements to demonstrate the program. Source code should be submitted and demonstrations completed by 5:00 PM, Monday, December 11, 2017.

The presentation

Each group will present a summary of their work to the class. Include background information on your topic, the motivation for your project, a description of what you did, and a summary of what you learned. Software demonstrations may also be appropriate. Prepare slides or web pages. Rehearse your presentation, paying special attention to timing. Given how busy everyone will be at the end of the semester, our schedule will be tight, so groups will not be allowed to run over the allotted time. Presentations will take place during a class minisymposium on 9:50-11:15 AM, Tuesday/Thursday, December 5/7, 2017. All group members must participate in the presentation. Attendance is required at the entire presentation session, not just when your group is speaking.

Grading

This final project accounts for 15% of the course grade. The grade will be based on all aspects of the project, including the proposal (5%), the progress report (5%) and annotated bibliography (5%), the design, documentation, style, and correctness of the software developed (10-40%), the content and writing style of the complete draft (10%) and final version (20-50%) of the paper, and quality of the presentation (15%). The breakdown between software development and the final paper will depend on the amount of programming required for each project. No credit is given for attendance at the minisymposium, but penalties will be applied for missing all or part of it.

Choosing a topic

Here are some ideas to consider when choosing your topic:

• Applying parallel computing to solve a problem in insert field of interest here. Ideally you would produce parallel implementations of a program to solve problems of interest to you. You might implement both with message passing and threading (or a combination) and do some performance analysis.
• Other parallel programming paradigms, like map-reduce.
• Building and configuring a small parallel computer (like a cluster of Raspberry Pis) and demonstrating its use.
• Demonstrating parallel implementations of algorithms using the METAL graph data.
• Implementations of algorithm visualizations in METAL to illustrate parallel execution of algorithms, or of algorithms of interest to parallel computation such as graph partitioners.
• Parallel discrete event simulation
• Parallel programming languages
• Parallelizing compilers
• Actor model of parallel computation
• Hybrid programming models, such as combining MPI for inter-node communication with OpenMP for intra-node communication
• GPU parallel computation

These are just ideas, and you should not restrict yourself to this list. I hope everyone can find an appropriate topic that they find interesting. You might also find it useful to browse recent journals and conferences proceedings to get a feel for some current research projects in parallel computing. The Supercomputing conference is a top conference in the field, for example.

We may be able to provide some hardward to each of a few groups who wish to pursue a project that requires it. If you think you may be interested in this, be sure to ask about it right away.

Academic Honesty Guidelines

Collaboration within a group is unrestricted. Since each group is working on a different project, you are free to discuss your projects with each other. If you wish to use or refer to any software libraries or outside source code beyond the standard language (C, C++, Java) libraries, check with me first. All sources must be cited properly. If in doubt about anything related to Academic Honesty, ask now and avoid problems later!

Final Thoughts

You have several weeks, so the expectation is for several weeks of work. You will not be able to do a good job if you put it off. I don't expect a Ph.D. thesis, but I do expect much more than your average lab assignment.