Attention Students:

Attention students who want to be my graduate research assistant:

Yes, I do have room in my research group for another graduate research assistant. However, I am interested in students who are considering doing a PhD at Maryland, and I can only keep track of students who are working in my research area (computer architecture, in particular systems architecture and memory systems). See the rest of my webpage for details and research reports on these topics.

As a rule, I do not fund students until I have had an opportunity to work with them on a semester-long project. If you are interested, I encourage you to drop by my office once you have arrived at the University of Maryland, and we can discuss project ideas. If you do well on the research project and demonstrate that you are independent, self-motivated, and capable of doing excellent research, then I will gladly fund you. I cannot offer any form of aid whatsoever to students before they actually arrive at Maryland or before I have worked with them on a research project.

Attention UMCP students who want me to be a reader on their Scholarly Paper:

I believe that the Scholarly Paper should be the equivalent of a Masters Thesis, without the novel research component. That is, it should present a thorough overview of a relatively large field (i.e. branch prediction is okay but branch prediction in the Intel x86 architecture is not), it should motivate why the field is important, and it should put all relevant research in perspective. In addition, the motivated student should also present a few of the open research problems in the field. The Scholarly Paper is not, in my opinion, a book report -- it is not merely a summary of a half-dozen papers. It should demonstrate that you understand a topic thoroughly, not that you have read a half dozen papers. Whether it contains a research component or not, it must have this significant overview component.

If you want me to read your Scholarly Paper, be forewarned that these are my expectations, and that I will not accept any paper that fails to live up to these criteria.

More on the topic (an example of a good overview).

How to do good research:

Research is the art of asking questions. It is very hard to ask good questions. If you ask questions the right way, how to answer them will be obvious. Time and experience will give you increasingly clearer perspective on how to ask questions well, but in the mean time, you should listen to your advisor (who has more of both under his/her belt). Here are a few brief pointers, which are mostly targeted to Computer Engineering:

How to write a good paper:

Writing well is something that requires a lot of practice and making a lot of mistakes along the way. Keep practicing all the time -- keep writing papers, as many as you can, even if you do not send them anywhere. The more you write, the better you will become at writing. Ask others to read your work and critique how well you get your point across, and make sure you listen to what they tell you. Here are a few ideas, but this is just a starting point.

The following is an outline that can be used for any research paper:

This outline should be used to write EVERYTHING in your paper: this should represent your ABSTRACT (each bullet would be one or more sentences); this should represent your INTRODUCTION SECTION (each bullet would be one or more paragraphs); and this should represent your ENTIRE PAPER (each bullet would be one or more sections). If you are writing a thesis, each bullet should be one or more chapters.

When discussing in detail (i.e. not in the abstract) how you plan to solve the problem/study the phenomenon/measure the performance behavior, it is extremely important to refer to previous work, describing what has already been done in the area and how it relates to what you are doing.

How to give a good presentation:

Presenting well is like writing well -- you have to practice to improve. So pull together a bunch of slides on the topic (start well in advance of the day that you present), and ask your colleagues to critique the talk. Improve the talk, based on their feedback, and then give it again. A hint: food is often a good bribe to get your fellow students to take an hour out of their day to critique your talk. Especially effective: beer, cookies, chips and salsa, home-made brownies, pizza, etc. Here are a few pointers: