I set up my chaintech av710 a year ago using the hi-quality rear outs which is automatically set to 96khz sample rate. I have been using foobar w/ kernel streaming and ssrc resampler set to 96khz. I am about to move the card into another computer and reformat and am trying to decide on the best way to configure it. I understand that quality cannot be gained from upsampling 44.1 > 96 but if I use the regular speaker output (not the high sample rate rear output) I gain the ability to set the sample rate to 44.1 but also lose the use of the higher quality wolfson dac. It seems that using the wolfson and upsampling gives better results but by the time i change all the settings in the control panel, then foobar, and then fiddle around w/ connections in the rear of the computer, placebo can start setting in. Any thoughts on this?

Also, I've read that it is recommended to set the output format to '24 bit padded to 32 bit' (when using high quality 2 ch mode). Why is this? Since the wolfson supposedly resamples everything to 24/96, why not just set it to 24 bit fixed point?

Thanks.

I've heard the same thing said about Creative's Audigy cards. I would like to know too.

My Terratec manual says something like this (from memory):
PCI bus transfers are 32bit. If you send 24bit size samples, they will be repacked so that four samples go in three packets. This may cause increased latency.

So it's not a quality issue.