>With an mmap-ed audio device, [I imagine] you would have the actual audio 
>DMA buffer in your user space.  This allows the Doom game engine to just 
>place the audio that needs to be played directly into the buffer, thus 
>saving the fork-ed soundserver process.

Hmmm.

While this sounds kinda interesting, it does present some problems.  For
example, how do you know where the sound card is reading stuff out of the
DMA buffer?  You could put a chunk of sound right into the middle of the
buffer pointer, which would sound really bad.

What we really need is the sort of driver abstraction for the capabilities
that the GUS provides.  With the GUS you can download samples to it and
just tell the card "play sample that starts here and ends here, and play it
over and over", and that sort of thing.  I'm not sure if it would make sense
to put that sort of logic in the sound driver, but it would be cool if you
could.

--Ken