The profiler is optimized for batch programs, and is especially useful with batch programs that run large numbers of transactions. It is more difficult to get good information from interactive programs. If user wait times are the issue you're trying to solve, trace files are more likely to return useful information than the profiler.
When you prepare to use the profiler, you should make an effort to run your application as cleanly as possible. This means making sure that your system isn't overloaded with large numbers of users, heavy system traffic, and so on. The cleaner the run, the more useful the information returned by the profiler.
The following steps describe how to perform profiling using default profiler and acuprof behavior.
You must compile with at least the -Gy option (to include at least minimal symbol information in the object file) for the profile to contain paragraph information. It is preferable to compile for full symbol information (-Gs) or for full source debugging (-Gd).
If a file called acumon1.xml already exists in the program directory, the profiler automatically changes the file name to create a file called acumon2.xml, acumon3.xml, and so on. This is intended to make it easier to compare multiple profiles of the same program.
Note that the automatic naming scheme uses the first unused number when naming the file. This means that if a directory contains files called acumon1.xml, acumon2.xml, and acumon6.xml, the next profile created in that directory is called acumon3.xml.
By default, acuprof creates a report file called acumon.rpt in the execution directory, then displays a message to indicate that the report was created successfully.