Quote:

I set a filter to only monitor RealBand.exe. I opened the program loaded a song hit play, stopped the song, saved, replayed, stopped, closed the program.
1,900,000 Events!!!

I'm not a systems engineer, and maybe someone who is could provide some guidance, but as I say, it's very granular.

However: I just set the filter to Process Name is RealBand.exe and Event Class is "Process", cleared the trace, started RB, loaded a .seq file, played a few bars, stopped it, closed RB, and in the status bar of ProcMon (lower left) it shows 151 of yes, a *very* large number of events. So I suspect you were reading the *total* events rather than those attributable to just RB. No?

The filter will provide a *lot* of selectivity. Examples: limit to file I/O events, registry events, etc. etc. You can also limit by many variables. Note particularly the "operations," where you can select for any combination of very specific events. (I would have thought the process tree feature will show all processes that flow out of RB as a parent, but haven't figured out yet how to make it do that.) You might actually want to be careful about limiting the process only to RealBand, since the thread it starts could be calling other processes which might be involved in the problem. One thing you could try is, without excluding any process, before firing up RB, letting the trace run for 10 minutes, and excluding any process that appears. "Filtering out the background" sort of thing. (I think you said you have a relatively clean environment, no internet, etc., so there shouldn't be that much. At least, this'll show you how "clean" it is. Mine is dirty as hell.) Then clear it and start RB.

Sorry I can't offer a sure fire approach. As I say, maybe the systems guys can. But you've got two sequences of (software induced) events one of which gives 100% success, one 100% failure. There's gotta be a filter that will highlight distinctions between them. Therein should be the clue.

Good luck, -Ron

EDIT - Well of course there'll be differences between the two sequences of events. By definition. I'm just thinking that somewhere in those differences should be the clue to the video stutter.

Last edited by rkl122; 09/20/09 11:54 AM.