Analyzing DevHealth Runs Easily

This is documentation of a tool created in November 2008.

Our team became fairly dependent on analyzing DevHealth runs when debugging memory issues.  We would run DevHealth automatically every few minutes as our applications were used for a longer period of time.  Tracking memory usage for a handful of processes over an hour so became extremely tedious and time consuming.

I wrote a small application to parse through any number of successive DevHealth outputs and produce a csv file, a row for every executable and the memory usage in bytes as the columns.  It became so easy to deploy new code, use our software for an hour or so, and within minutes have a graph of memory usage for every process we cared about.  We tracked our apps in addition to gwes (graphics, windowing, and events subsystem) and our database engine.

I will post the source soon.  I originally wrote it as an in-house exec where we had to specify the apps to watch at design time.

Memory usage was calculated as the following: heap usage, code usage (non-ROM), r/o data, r/w data (page, stack, etc.) In DevHealth symbols with a 4096 byte page size:
(r/w(r, rw, E, D, H) + c + R + p + S) * 4096