An important part of this course is reading research papers. There are two reasons for this: (1) only by reading resesarch papers will you thoroughly understand many of the algorithms and techniques that we will talk about in class and (2) reading research papers properly takes practice, and is an important skill to develop, especially for those of you that go on to a research career.
Schedule will be posted here as the class proceeds. PAPERS ARE DUE BEFORE CLASS
Papers are available on Piazza
|1||Jan 24||M. Gort and J. H. Anderson. Accelerating FPGA Routing Through Parallelization and Engineering Enhancements. IEEE TCAD, 31(1):61–74, 2012|
|2||Jan 31||M. An, G. Steffen and V. Betz, “Speeding Up FPGA Placement: Parallel Algorithms and Methods,” FCCM, 2014.|
|3||Feb 14||Wenyi Feng, Jonathan Greene, Kristofer Vorwerk, Val Pevzner, Arun Kundu, “Rent’s Rule Based FPGA Packing for Routability Optimization”, FPGA, 2014.|
Submission: Papers should be submitted to the instructor using a private message on Piazza. When you create a message on Piazza you can choose a folder for the message. Chose the correct “paper#” folder based on the numbering above.
Your review should be short (1 page seems reasonable, two pages is too long).
In writing the review, you should imagine that this paper has been submitted to a conference, and you are reviewing it to determine whether the paper should be accepted. The review will have the following five sections. Label each section clearly.
1. Summary of the main contributions of the paper: This should be more than a re-statement of the abstract. What do you think are the important contributions of the paper? This section must be at most four sentences.
2. Things you like or dislike about this paper: This is the bulk of your paper review (between ½ and 1 page). You may comment on things like: (not all in one review):
Back up your statements, and try to avoid generic comments that could apply to any paper. Statements like “This is an important problem” are not very useful, however, a statement that outlines why the problem being solved is important (relative to previous work) or why the authors are tacking the wrong sub-problem is much more useful. Re-stating the author’s comments on the value of the paper is also not useful; every author thinks their paper is important, but that is not always the case. Note that, in many cases, you may need to read some of the previous work referenced in the paper to come up with a meaningful review.
3. (At least ) one positive point about the paper: Even a really bad paper has at least one redeeming quality. Identify one positive point about the paper, even if you don't like it.
4. (At least) one negative point about the paper: Even a really good paper has some shortcomings or limitations. Identify at least one.
5. The bottom line: Give this paper a score between 1 and 6. 1 means the paper is horrible and should never be accepted to a conference. 6 means the paper is groundbreaking (or was, when it was published). Try to avoid middle scores (2 or 3) when possible. It may seem that the assignment of a number is a bit arbitrary, but that is what happens when papers are submitted to conferences. You don't have to justify your score in this section, since your other sections above (primarily Section 2) will hopefully provide enough jusification. No waffling – you have to give a number.