I\'m getting weird audio breakups when i get to a high voice count w/ giga. i previously thought it was a problem with my new IDE controller, but i think i\'ve eliminated that as the cause. I installed a drive on the secondary IDE channel with nothing else on it but a few .gigs to test, and i still get the same problem, although at a higher voice count. same thing with the C drive, but at a lower voice count (104).
THE PROBLEM: when i get to a certain voice threshold (say 120 voices) the new notes cease sounding, and previously played notes begin \"winking\" in and out. for lack of a better description, it sounds like the drive is trying to keep up with streaming all the voices, and can only keep a few going at any given time. this gradually improves as the voices peter out. the sound is not unlike some excessively \"modern\" composition (think 12-tone), and does not seem like the sound of voices being dropped. there is also a weird \"electronic\" edge to the sound. interestingly, the voice meter freezes at the point where this happens, and won\'t indicate more voices sounding (despite more keys being pressed).
i have 2 other giga systems that do not have this problem, even when maxing out their polyphony.
SOLUTIONS I\'VE TRIED: AFAIK, i have applied all the recommended performance tweaks (UDMA, etc). i believe i have eliminated possible IRQ conflicts, but i\'m still moving things around trying to cover all the variables.
DOES ANYBODY HAVE ANY IDEA WHAT THE PROBLEM COULD BE?!?!?
i would greatly appreciate anyone\'s suggestions, especially if this sounds like something you\'ve experience and solved.
frighteningly, the next day my reinstalled win98se corrupted the data on my sample drive, so i must have installed some windows update in my attempts to fix that, that caused the bouncy sound to reappear. fortunately, it went away again after reinstalling windows again.
as of the end of yesterday, everything was ok-i\'m sorta scared to check it again today!
i\'m glad i was able to post some sort of solution-i found several references to this in the archives, but no solutions.
anyway, thanks for taking the time to give me your thoughts.