With the computer AND a soundcard, there are basically two processors. Your CPU can do what ever it wants as long as the software tells it to do the right thing. Then you have your soundcard with its DSP processor on board that is limited to audio only related processing with limited resources.
In my thoughts, to maintain the ultimate SQ, this would be ideal:
All sound adjusting is done by mathematics before the soundcard ever gets a hold of the signal
- Resampling 44.1kHz audio to 48kHz/96kHz (if needed depending on what the soundcard uses to process)
- EQing (via VST plugins)
- Time delays (Freq Allocator)
- Crossovers (Freq Allocator)
-The CPU will probably be doing ALOT of work.
Option 2 (If you have a kx driver compatible card)
Handles things the soundcard cannot do or do well
-resampling 44.1kHz to 48kHz or 96kHz depending on soundcard used
-EQing (via VST plugins)
KX driver compatible Soundcard
-Time delays (Kx driver w/UFX plugin package)
-Crossover (Kx driver w/UFX plugin package)
The workload is split between the two processors
It's not Bit perfect, but CD audio can be refined or smoothed out by resampling and not really affecting quality. The D/A dacs are really the only thing you need to worry about with SQ with both of the above scenarios. At least this is what I am thinking...