[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Getting Elapsed Times

In several messages and followups, Wong Joo Fung wonders how to
collect information on the actual elapsed time. Here are some
guidelines I developed for running an experiment.
1. force a garbage collection before the trial begins
2. turn on egc (egc t) since this will tradeoff small delays
   for the larger delay of a full gc,
3. don't sample too much (try 10 times per seond or less if you're
   recording mouse position information.
4. use fast I/O packages (i.e. the low level routines) to write out
   data and preferably store the data in memory (e.g. an array)
   during the trial and write it out at the end of the trial.
5. Use the monitoring tool from Mark Katorwitz at Carnegie Mellon
   University if you'r really concerned about the inefficiencies
   in your code. Just make sure that the cpu time is far less than
   the actual elapsed time. You may also need to run some critical
   parts of the code with interrupts disabled.

Hope this helps.