ANNOUNCE: HEAD is open for development again

[Matthias Welwarsky]

On Sunday, 15. August 2004 21:51, Charles Samuels wrote:
> On Sunday 15 August 2004 4:12 am, Matthias Welwarsky wrote:
> So if your sound card supports 44.1kHz, then set the kernel to 44.1kHz. 
> The point is that anyone who needs complete and utter quality will either
> be using a good soundcard (with multiple inputs), or will know they need
> the quality, won't be using Linux on the desktop, and will turn this off
> anyway. The point is that we're not targeting these users anyway.

I don't really see the need to break it by design, beforehand :)

> A linear resampler won't distort as much as the mp3 codec itself!

OK, but there are lossless codecs, too.

> So once you can get a scheduler that can guarantee the the latency of
> user-space-mixed audio, then maybe you'd have a point.

Well, you will basically never get hard realtime anyway, and even if you mix 
in kernel, the feeding processes are still in userspace and thus at mercy of 
the scheduler. Doesn't gain you much if only the mixer is in kernel.

But, honestly, how much of a problem is latency for music playback? It is a 
problem with video playback, admitted, but latency is just a function of 
buffer size, so you should be able to compute the delay and calibrate your 
playback.

And even for video: You hardly notice two or three frames desync, and even 
with NTSC video material this computes to 99ms of acceptable desync, which is 
a _lot_. And if your backend, be it a soundserver or /dev/dsp is able to 
report the added delay..

Action games have insanely high requirements for A/V delay, to the point where 
they directly map the driver buffers and mix their effects directly in the 
in-kernel memory. Still, their mixer process runs in userspace.

-- 
Matthias Welwarsky
Fachschaft Informatik FH Darmstadt
Email: matze at stud.fbi.fh-darmstadt.de

"all software sucks equally, but some software is more equal"
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: signature
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20040816/9a4f20aa/attachment.sig>