Stop searching, you are confusing two different things.The setting in Vista/7 just defines the preference for output when mixing, (which is done in legacy mme, directsound, and even wasapi in shared mode).In your screenshot, a wav file ripped from a CD would play like this(Player)Wav 44.1Khz 16bits -> (mixer) increase bit to 32bit, resampling to 96Khz, reduce bits to 24bits -> (Driver)reproduce. Generally it would not resample, but that really depends on the manufacturer.In windows XP, the setup is:(Player)Wav 44.1Khz 16bits -> (mixer) resample to a samplerate that the driver supports (this meant 48Khz with AC-97 soundcards) -> (Driver) reproduce. Some drivers used internal resampling on the hardware, others used software resampling in the drivers, and others used the resampler of Windows, which could be set to different qualities.(For a 24bits one, if the driver supports it, the same path applies, so it is automatically 24bits).In my Windows 7, I have the setting to 44.1Khz, 24bits, since i usually play music sampled at 44Khz, and the players might output at 16bits, 24 or 32bit, so I set to 24bits so that i don't degrade the higher bit-depth ones.It does not improve at all the 16bit signals (except, maybe if the hardware or drivers do something different depending on the bitdepth).
In XP, if you are not playing any other sound and the application that you use allows to output at 24bits, then, the audio will pass as is (except if it requires resampling at the driver's end).In other words, in XP it is more dependant on the application to use the correct output method (for example, its directsound to support 24bits) than to an OS setting.