Re: overrun with 'arecord' - why ?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, 8 Apr 2011 09:48:04 -0700 (PDT)
Bill Unruh <unruh@xxxxxxxxxxxxxx> wrote:

> On Fri, 8 Apr 2011, Sergei Steshenko wrote:
> 
> > Hello,
> >
> > I've tried to run 'arecord' as part of simultaneous playback + capture "rig"
> > (for acoustic measurements) and noticed overruns.
> >
> > So, even plain single 'record' occasionally produces overruns:
> >
> > arecord -D hw:0,2,0 -c 2 -r 96000 -d 6 -f S32_LE recorded.wav
> > Recording WAVE 'recorded.wav' : Signed 32 bit Little Endian, Rate 96000 Hz, Stereo
> > overrun!!! (at least 855507586.521 ms long)
> 
> Wow. That is a three  month long overrun! Or perhaps it is a bug in arecord. 
> You are using 96000 why? Remember that you are pumping out about 1MB/s. What
> is the cable-- USB cannot handle that. I would suspect something spurious is
> going on. (PS-- you are using 32 bit recording. You do know that there is
> absolutely nothing in the world that has a signal to noise ration of 180dB. In
> fact I suspect that you are not getting anymore than 16 bits of useful
> infomation per recording session. (S/N of even 90dB is pretty rare).
> 
> 
> > .
> >
> > So, my question is: "Why ?". It's a 2.6Ghz machine with SATA disks. Two cores,
> > web browsers are the most active tasks (nothing fancy, no sound activity on
> > the side of the web browsers). Effectively one core is free.
> >
> > Any ideas ?
> >
> > The gear:
> >
> > Card: HDA NVidia
> > Chip: Realtek ALC883
> > .
> >
> > Thanks,
> >  Sergei.
> >


The sound card is on-board (i.e. on-motherboard), i.e. no USB whatsoever.

Sustained write disk speed on my machine is no less than 20MBytes/second - I
remember such number running 'scp' with all its encryption/decryption
overhead.

I've seen messages like

"at least 855507586.521 ms long"

for years, they happen with underruns too (e.g. in case of an MP3 stream is
played and the connection is unreliable). I first noticed them using
'ecasound' and Kai Vehmanen told they were an ALSA bug - now that I see
'arecord' producing them I very much believe Kai Vehmanen.

The sound card _does_ support 24 bits, but its ALSA driver doesn't, i.e.
my choices with the card are 16 and 32 bits.

Being an electronic engineer who also happens to be the sole developer of
my own measurement rig I definitely know from my observations with
oscilloscope that at some frequencies my output level will be much (tens of
dbs) below nominal, so I'd like to use the card's full dynamic range.

The rig has its frequency plot very far from flat (for _acoustic_/physical
reasons) and it's the way it should be.

As people have already realized, 24 bits is the recording format which allows
looseness in level settings - the only concern is to have enough headroom to
avoid clipping, and 16 bits is the delivery format.

Oh, and there are no xruns using 'aplay' with the same bandwidth. In reality
I first start 'aplay' and then 'arecord', and for several seconds they run
in parallel. The overrun problem this thread is about happens with _just_ 'arecord'. I.e. I've started the thread making sure the problem is an
'arecord' problem with no other sound activity.

Regards,
  Sergei.

------------------------------------------------------------------------------
Xperia(TM) PLAY
It's a major breakthrough. An authentic gaming
smartphone on the nation's most reliable network.
And it wants your games.
http://p.sf.net/sfu/verizon-sfdev
_______________________________________________
Alsa-user mailing list
Alsa-user@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/alsa-user


[Index of Archives]     [ALSA Devel]     [Linux Audio Users]     [Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite News]     [Yosemite Photos]     [KDE Users]     [Fedora Tools]

  Powered by Linux