Lab 4 - Performance Lab: Optimizing Program Performance

The files you need are available in directory /afs/cs.cmu.edu/academic/class/15213-f02/L4.

You will use our benchmark server both to test your code and to handin your final solutions. Here's how it works:

• Register yourself as a group using the Group Registration link. You should use your Andrew ID as your Group ID, since that's the only way we'll be able to identify your submission. You may choose any "moniker" you like.
• Once you are registered, you can submit either of the two files: poly.c and rowcol.c by clicking on the Benchmark server link. It is important that you use these two names for your files, since that is the only way the server knows what to do with your file.
• The performance statistics link lets you see how you are doing and compares it to others in the class. By clicking on the appropriate column heading, you can get the results sorted according to the scores for any of the four measurements.
• The handin history link lets you see a log of all of your submissions.

Resources

• Group registration
• Benchmark server
• Performance statistics
• Performance statistics for new tests
• Handin history

Notes

• There are various ways you can fool our testing code and our benchmarking code on this assignment. For example, its possible to get a negative CPE measurement by doing wasted work on smaller polynomials.

  Instead of making up a bunch or rules and modifying our testing procedures, we'll just look at your solutions for weird stuff when we grade them, and we'll invalidate those solutions. As a general rule, your code should compute the correct results for arbitrary data and it should not do any extra computation solely for the purpose of slowing things down. If you're not sure whether your optimizations are legitimate, check with us.