[Qemu-devel] Audio

[malc <av1474@comtv.ru>]

Hello,

At http://www.boblycat.org/~malc/code/patches/qemu/ you will find latest
(21_aqemu) audio patch.

New things relative to previous one:

  1. ALSA failure to set the format is no longer considered to be a
     hard error, in this situation ALSA is queried for the format it
     deems acceptable and work with that (On this PPC Linux box ALSA
     always returns EINVAL when asked for S16_LE but merrily works with
     the default - S16_BE)

  2. Added support for 32bit per sample formats

  3. Intel Controller Hub N Function 7 AKA AC'97

New things relative to the CVS:

   1. All of the above

   2. EsounD driver by Frederick Reeve
      (If someone wants aRts driver (s)he is welcome to do it, the task
       is mostly renaming the identifiers and linking with libartsc)

Now to some mildly amusing performance observations. [1]

QEMU running Slackware 11.0 (Kernel 2.4.33) on PPC 7447A 1.3Ghz
demonstrated that:

i810_audio [2] is faster than es1370
snd-intel810 is on par with snd-ens1370

i810_audio and es1370 blow snd-intel810 and snd-ens1370 right out of
th water.

The difference is toned down on Athlon 1G - where QEMU has a luxory of
using KQEMU, but nevertheless.

Perhaps the situation is improved in latest ALSA incrarnations, i don't
know.

[1] Observations courtesy http://www.boblycat.org/~malc/apc/

[2] i810_audio might, and usually does, require setting the `clocking'
     parameter upon module installation

-- 
vale