Ok, I messed up HW(audio&system) "real" clocks and SW(renderer) generated clocks. Thanks for the explanation. If I think more that HW point of view there sure is different oscillator for audio.I'm not really familiar with different clocks there. You mean that my mb's (or ATIs) audio chip clock might not be genereted from the system clock?It might be that your PCs system clock is running on a different speed that the audio clock. That would explain the small drifting since currently the audio HW clock / sample comsumption is not used to generate the reference clock. There is currently some developent effort spent on that (not by me) and hopefully we can have a "real" audio HW based reference clock at some point.
Audio HW clock is not ever (at least I think it is not) generated by the system clock - audio HW has it's own clock source that is used to time the audio renderering.
What I ment was that the audio renderer itself is not currently generating the reference clock (used by rest of the directshow graph - liek EVR presenter) based on the audio HW clock. This is something that needs to be fixed at some point in the development. There should be currently one non-MP developer looking into this "feature".
So with that PC I really can only wait until MP audio renderer is based its clock on the audio HW clock.
Finland