Microbenchmarking Issues
For your "presentation" on your microbenchmarking paper, please
address the following issues. If you cannot find a direct answer in
the paper, please use your judgement and make your best guess. You
should indicate (in some humble manner) when you coming up with things
beyond what was in the paper.
Of course, the order in which you present these issues and the level
of detail you go into, is up to you.
- What information do they (want to) acquire?
- Why is this information useful?
- Do they omit gathering any useful/related information from their
system-under-test?
- How does their microbenchmark work?
- How long does their microbenchmark take to run? Is the runtime a
function of anything?
- What assumptions do they make about the system-under-test?
- Going beyond the basics... Do they require an otherwise quiet
system? What would happen if the system weren't quiet? Would it be
possible to acquire this inforamtion in a system by passively
observing existing operations instead of inserting new operations?