Department of Computer Science and Technology

Course pages 2017–18

Chip Multiprocessors

Assessment consists of four short essays, a presentation and a longer survey paper. The marks given for each are listed below:

  • 60% - Four essays discussing the reading assignments (e.g. papers and reports)
  • 10% - A 20 minute presentation on a specific paper or area
  • 10% - Participation during the module's reading group discussions
  • 20% - A longer survey paper

Important Dates:

  • Essays discussing particular papers/reports should be submitted before the seminar at which they are discussed
  • The survey paper is due by the first day of the Lent term
  • The date of your presentation will be scheduled by the module convener

Presentations

Participants will give a short presentation during one of the seminars. The talk will introduce and discuss a paper, parallel programming language or interesting processor design. The topic should be discussed with and confirmed by the module convener. The talk should be approximately 20 minutes long. Speakers are expected to use a small number of slides.

"How to give a good research talk", Simon Peyton Jones

Reading Group

Participants should prepare for the reading group component of the course by reading and making notes on the assigned reading.

Essays

Participants are asked to write four essays in total. Each essay should summarise and discuss a paper or theme from a week's reading assignment. group. The essays should be completed and submitted before the reading group session. The suggested length of each review is around 1,500 words.

A suggested structure for essays is as follows:

  • A summary of the paper.
  • Background. Clarify any important concepts required to understand the work and perhaps identify and briefly discuss other important work in this area.
  • Contributions. State what contributions you think the work makes. Is there an additional contribution that the authors could have made that would have been interesting? Have they clearly identified their contributions?
  • Further comments. Make any further comments about the work, e.g.
    • Do you think the work is novel?
    • What are the limitations and weaknesses of the work?
    • How does it relate to other work in the field?
    • Can you see any flaws in their experimental methodology or approach?
    • Can the contribution be applied generally or only in very special cases?
    • Are there outstanding issues that would be good to address?
    • What were the best and worst things about the paper?
  • Conclusions. Summarise your thoughts on the paper.
  • 3-4 interesting ideas or questions to help stimulate discussion

The following resources may also be useful:
Prof. Dodgson's lectures on reading and reviewing papers
"Writing reviews for systems conferences", Timothy Roscoe, ETH Zurich
"How to Read a Paper", S. Keshav, University of Waterloo

Survey Paper

The survey paper should be 2,500-3,500 words in length. The paper should provide a clear description of a particular problem or field and describe why it is significant. It should provide any necessary background and perhaps a short history of the field. The paper should provide a description of the key research challenges. It should go on to discuss and contrast the papers you have chosen to discuss. Try to avoid simply discussing each paper in turn. It is often better to use references to key papers to illustrate ideas and trends. Classifying approaches is also often useful. Take care to discuss the relative advantages and disadvantages of different proposals each paper presents. Finally, a discussion of the remaining research challenges or fundamental barriers to further progress would be of interest.