Hi Dennis, [More processing power doesn't proportionally lower latency :] >I am running a G3 400/200 accellerator card in a Mac 7600 with a >K1212 I/O, I can monitor inputs from the card and have no >problems with latency at all. The 1212 must have a monitoring path built-in to the hardware. This would not involve the CPU, so latency is very low, probably just an A/D/A conversion. >I can understand latency in realtime fader adjustments but one >aspect of the whole latency issue I don't understand is why >programmed fader changes should have any latency at all. >Wouldn't it be a relatively simple matter (and obviously meritorious) >to program logic so that it can look ahead to fader movements and >make whatever adjustments are neccessary according to the latency >of the system it is running on? The automation wasn't initially designed to look ahead. But it obviously can & should do so. >43ms is awful but again there is nothing at all like that with the >K1212. Maybe the drivers still need a little refinement. So how quickly does it respond to f.e. an effect or EQ being switched in or out with the bypass button? That's latency, and you've probably got as much or more than the MOTU users. In fact it probably applies to everything you can do except monitoring, and maybe start play... Logic can prepare audio when stopped, to minimize the delay. >>How cool would it be to put mixing/automation on one G4 >>processor and plugins on another, etc, etc. >My understanding of the G4 velocity engine is that it does almost >exactly what the person above is describing except that it does not >even need two cards to process in parallel. It can for example >process four 32bit streams at once. I think it does one process 4 samples at a time, not four different processes. These 4 samples could be separate streams, but are more likely to be multiple samples from a single stream. Cheers, Thomas