Reading through all this discussion, I think most people do not consider enough the interaction between hardware and software. The key to good software for audio playback is to have a constant but low load on the hardware whether it be CPU or memory.
When the software memory and CPU load is high, this excites interaction of hardware circuits and power supply, what happens is unpredictable because it depends on hardware design, which could effect clock jitter for digital circuits, or couple into analog power supply.
Another approach would be to have short but heavy load of software operation then have a VERY low load for long periods in between.
No matter what method used, basically as long as there is no bit drop throughout the process, players should make no difference; it's bad hardware that produces the audible difference.
If there is no audible difference, the issue is more complicated. But generally it just means that there are more dominant problems in the system that creates a masking effect.
