Adminsitrivia

Syllabus

  • Programming languages: Global address space languages, stream-based computing.
  • Alternative programming models, including run time support.
  • General-Purpose computation on GPUs: IBM Cell, software techniques.
  • Performance programming: memory hierarchy management, overlap techniques.
  • High performance I/O: data servers, storage systems, remote I/O
  • Irregular problems.
  • Feature extraction and tracking.
  • Your favorite topics.

    • Required Background. You should be familiar with the principles and practice of parallel computation, which may be met by having taken CSE 260, CSE 160, or another course in parallel computation, or having the equivalent experience. Some of the topics will involve numerical algorithms, but the material will be self contained. Students are welcome from Departments outside Computer Science and Engineering. If you have a favorite project you'd like to work with, let me know.

    Course requirements

    Class participation and preparation (30%)
    Your grade will be based in part on your participation in lecture and your advance preparation. We will read about 2 papers for each lecture. Be prepared to ask and answer questions and to contribute to the discussions. You must also submit, before lecture, a 1 page write-up for each paper assigned that lecture. You must do this for 10 of the 20 lectures.

    In class presentation (10%)
    You’ll lead the discussion for the assigned reading into two different lectures. Be prepared to ask questions. I will moderate.

    Research Project class presentation (60%)
    Your class project will count toward 60% of your grade. A project may be done individually or in teams of two, and you should allow 6 or 7 weeks to complete it. See the Projects web page for further instructions, as there are four milestones with deadlines.

    Maintained by . Last modified: Sun Feb 4 21:34:54 PST 2007