• Home
  • Announcements
  • Course Spotlight
  • Course Projects

  • Schedule
  • Lectures
  • Recitations
  • Quizes
  • Recipes

  • Student Projects
  • Competition
  • Backgammon
  • Battery Model
  • Blue-Steel
  • more to come
  • Links
  • Sponsors
  • The project competition is scheduled for Thursday February 1 form 10am to 12pm in room 32-141. The project teams will persent in the following randomly chosen order.

    Global Illumination
    Battery Simulation
    Ray Tracer
    Software Radio
    Speech Synthesis
    Backgammon Tutor
    Molecular Dynamics
      

    Each presentation is limited to 15 minutes and should include the following:

    • An overview of the project goals
    • Relevant technical details for the algorithms you use
    • How you determined which parts of the algorithms to parallelize and accelerate with the SPEs
    • Your methodology for parallelizing the algorithms
    • The performance metrics you use for the evaluation
    • An evaluation of your parallelized implementation
    • A live demonstration of your application running on the PS3

    Project evaluations are based on the following criteria:

    • Performance. We will empirically determine how well your implementation harnesses the Cell SPEs to accelerate computation. We will use profiling tools to quantify performance, comparing execution time when using the PPE to the execution time when the SPEs are also used. The profiling will attempt to measure performance scalability as the number of available SPEs is increased from one to six. The performance analysis will be done after class.

    • Project Scope. You chose the projects, and as a result, no two projects are the same. We will determine the scope of the project based the algorithms you implement, as well as your parallelization approach.

    • Project Completeness. The projects you chose are ambitious and exciting. The time for implementation is short and limited. Your code should be robust. That means the behavior should be reproducible, and the code should not repeatedly crash, deadlock, or produce incorrect results.

    • Presentation and Demo. We strongly encourage you to take the time and rehearse your presentation at least once before the final competition. How you present your project will directly impact how we perceive your project in terms of its complexity and completeness.


    MIT
    Comments and questions to 6.189-chair@mit.edu