Re: ALSA drivers
From: Clemens Ladisch (clemens_at_ladisch.de)
Date: Tue, 29 Jun 2004 16:38:09 +0200
> For instance, for some reason I can't understand (and nobody else I have
> discussed the issue with can, either), the ALSA drivers don't seem
> to play well with the esd daemon in Gnome - invariably, one or more
> channels in the board get muted when launching Gnome with an ALSA-aware
> esd. This could be esd's fault, of course;
... and it probably is. It doesn't happen with other ALSA
(ALSA drivers mute everything when loaded, but your computer should be
configured to restore previously saved mixer settings after that.)
> To top it off, configuring the ALSA drivers is significantly more
> cumbersome than configuring the OSS ones;
This is the price one has to pay for more flexibility. And the
configuration should be done by your distribution's setup tool anyway.
> in 2.4.* kernels (where they must be compiled as modules)
... in 2.6.x kernels, too, ...
> they result in a bewildering number of modules loaded at run time.
Modularization is considered a feature (mostly for developers, as it
And you can unload any modules that you do not need.
> I can't help but wondering why they require at least 7 or 8
> modules, when the OSS drivers get the job done with a couple of
Because "the job" includes more things, such as a framework and
libraries for the hardware drivers, a complete MIDI sequencer, a
common API for applications, and, not to forget, OSS emulation.
> Now since they are being included in 2.6.* kernels instead of OSS drivers
In addition to OSS drivers.
> they must be somehow superior to the OSS ones. My question is, how
> so? I mean, from the point of view of an end user, not a
Mostly support for (the features of) professional-grade hardware.
This isn't a big issue for 'typical' desktop users.
Oh, and wavetable support on SoundBlaster cards.
> I am willing to believe that the code is better,
It is newer, and bigger. This implies that the number of bugs tends
to be higher ...
> offers a nicer API,
with horrible documentation
> and has more solid foundations. However, for me, an end user,
> right now, that is not relevant if they don't do at least as well
> as the OSS ones. Which they haven't.
Then use OSS. Nobody forces you to use ALSA.
(Well, there _are_ many applications that require ALSA, but this won't
affect you much unless you want to create music.)
> If they are so bloody superior to OSS, how can one get the ALSA
> drivers to deliver? Maybe I have been unlucky enough to try them
> on precisely the sound boards where the OSS drivers do better?
> ... (CMPCI, Ensoniq AudioPCI, ESS18xx, VIA 82Cxxx and some
Those are indeed the drivers with the most problems, although most of
those are caused by cheap hardware.
> Do people here have had similar experiences?
The drivers for my hardware (ICH5+AD1985, YMF754, UA-1A, SC-8820,
Bt878, OPTi601+AD1848+YMF278B) work just fine.