Subject: Re: [ecasound] recording problem (underrun)
From: justin (justinrosander_AT_access-4-free.com)
Date: Mon Jul 21 2003 - 01:38:06 EEST
Hey,
I don't know what Kai would say as far as your syntax goes, but you're using
two different alsa devices in your syntax. If your device is called
"alsa,in1" then use that consistently; right now you're using "alsa,in1" and
"alsa,default". My sound card's hardware "plug" name (see alsa documentation
for your soundcard) is "ens1371", so I like to use "alsa,ens1371" both for in
and output.
Also, as a practice for my own consistency, i list input chains first:
ecasound -r -b:256 \
-a:1 -i <arg>
-a:2,3 -i <arg>
-a:1 -o <arg>
-a:2,3 -o <arg>
Maybe I get that practice from learning HTML and XML well-formedness, but it
seems to give me consistent results, and those being good. Try cleaning up
your syntax first. Your system speed is fine; I've run ecasound on 100 mHz
Pentium, 32MB RAM, and had good results. Ecasound uses low CPU power, and I
can testify that with 6 reads and one write, I'm not even close to 30% CPU
usage on my 600 mHz CPU. It's a good thing you're not using X during record,
because it saves a LOT on the latency end of things, and you may want to try
a low-latency patch.
Another thing I like to do to "open up" the PCI bandwidth is:
setpci -s <your_soundcard_device> latency_timer=ff
Find your soundcard device by typing:
lspci -v
Your device will look like this:
00:0e.0 Multimedia audio controller: Ensoniq ES1371 [AudioPCI-97] (rev 08)
Subsystem: Micron Tazer
Flags: bus master, slow devsel, latency 32, IRQ 9
I/O ports at d800 [size=64]
Capabilities: <available only to root>
Use the numbers at the top on the left (mine being "00:0e.0")
I can't really vouch to say exactly how much this saves on the latency end of
things. I do know when an xrun occurred without it, JACK would report runs
of, say, 10.000 milliseconds. When I did "setpci", the xruns dropped to
about .300 milliseconds.
Hope this helps,
Blessings,
Justin
On Saturday 19 July 2003 05:10 pm, you wrote:
> Hello again!
> I got another problem. I did some realtime recording with multitracks. I
> called:
> ecasound -a:1,2 -i alsa,in1 -a:1 -o out1.wav -a:3 -i null
> -pn:metronome,120 -a:4 -i monitor.wav -a:2,3,4 -o alsa,default
> When I record I get occasional underruns or whatever. It pops now and
> then. Some system-info:
> 700Mhz cpu (don't know exactly) 128MB ram, no x-server (even killed
> stupid apps like cron, sshed, nscd, etc.), kernel 2.4.19 (NO lowlatency
> patch) I ran ecasound as root with realtime-scheduling.
> Any ideas why this would happen? What can I do about it?
> Kindest regards
> Julien
>
> Julien Patrick Claassen
> jclaassen_AT_gmx.de
> julien_AT_c-lab.de
> http://www.geocities.com/jjs_home
>
> SBS C-LAB
> Fuerstenallee 11
> 33102 Paderborn
>
> Phone: (+49) 5251 60 6060
> Fax: (+49) 5251 60 6065
>
> www.c-lab.de
-- "I drank what?" --Socrates
This archive was generated by hypermail 2b28 : Mon Jul 21 2003 - 01:27:18 EEST