From alsa-devel-owner@alsa.jcu.cz  Sat Mar 13 03:48:56 1999
Received: from executioner.lis.net.au (mail@executioner.lis.net.au [203.35.83.3])
	by marvin.jcu.cz (8.9.1a/8.9.1) with ESMTP id DAA03144
	for <alsa-devel@alsa.jcu.cz>; Sat, 13 Mar 1999 03:48:09 +0100
Received: from uucp by executioner.lis.net.au with local-rmail (Exim 2.05 #4)
	id 10LeT1-0006ot-00 (Debian); Sat, 13 Mar 1999 13:47:59 +1100
Received: from kerry by gotss.spice.net.au with local (Exim 1.92 #1)
	id 10LdxN-0000Xp-00 (Debian); Sat, 13 Mar 1999 13:15:17 +1100
Message-ID: <19990313131517.A2076@gotss.ml.org>
Date: Sat, 13 Mar 1999 13:15:17 +1100
From: Kerry Hoath <kerry@gotss.spice.net.au>
To: alsa-devel@alsa.jcu.cz
Subject: Re: Enlarging the dma buffer for audiopci
References: <48A7CB4095A7D11187EF00609748374C02F161AF@w1sxms01.mobilkom.at>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
In-Reply-To: <48A7CB4095A7D11187EF00609748374C02F161AF@w1sxms01.mobilkom.at>; from T.Mertes@mobilkom.at on Fri, Mar 12, 1999 at 03:18:34PM +0100
Reply-To: alsa-devel@alsa.jcu.cz
Sender: alsa-devel-owner@alsa.jcu.cz
Precedence: list

If your system has IDE disks try ide interupt unmasking with hdparm and enabling
multiple sector mode. Often "load" is the system reading and writing lots of
stuff to and from disks. If you unmask the ide nterrupts you can get better
interrupt latency for the sound drivers, and you can aslo have better system
responsiveness. I use
hdparm -u1 -m8 /dev/hda
on my ide systems and I often increase sound interrupt priority with irqtune.
Regards, Kerry.
On Fri, Mar 12, 1999 at 03:18:34PM +0100, T.Mertes@mobilkom.at wrote:
> Hello
> 
> I have a SB 128 PCI sound card that i am using with
> linux 2.2.3 and the ALSA sound driver "audiopci".
> 
> I have sound dropouts when playing wave files under load.
> It comes from the fact that the program which plays
> (sox, play, ...) has to wait sometimes (because of the
> load) at the read() for too long. While the program waits, the
> 128kB buffer of the driver runs out and there is a dropout.
> 
> This seems to be really a buffer problem, because I have it
> also with the sound driver from the kernel and with the
> SB 16 vibra sound card.
> 
> I know that there are other ways to solve the problem.
> (sound demon/playing program which use threads and
> big buffers.) But I want first to investigate possibilitys to
> repair the problem on the device driver side.
> 
> There are two possibilities here:
> 
>   1. The sound driver does not use its 128kB efficient
>       (After issuing a write() the buffer should always be
>       almost full). I do not think that this is the case,
>       but I would be happy if the driver writer would check
>       this also.
> 
>   2. A buffer of 128kB is too small (I think this is the case).
>       I would like to enlarge the buffer (I have to try out,
>       which buffer size is necessary). The current audiopci
>       driver limits the size of the dma buffer to at most 128k.
> 
> My question is:
> 
>   - Is it possible to enlarge the buffer of the audiopci driver
>     to even several megabytes (configurabel like it is now).
>   - Is there a built in ALSA or PCI-DMA limit which
>     prohibits it? Can we lift this limit?
>   - What is necessary to do to enlarge the buffer
>     (a patch would be great :-) ).
> 
> 
> Greetings Thomas
> 
> ------
> To unsubscribe from <alsa-devel@alsa.jcu.cz> mailing list send message
> 'unsubscribe' in the body of message to <alsa-devel-request@alsa.jcu.cz>.
> 

-- 
--
Kerry Hoath:
kerry@gotss.spice.net.au or khoath@lis.net.au
ICQ UIN: 8226547


