From alsa-devel-owner@alsa.jcu.cz  Tue Jul 14 12:00:29 1998
Received: from iem.mhsg.ac.at (root@iem.mhsg.ac.at [193.170.129.55])
	by marvin.jcu.cz (8.8.8/8.8.8) with ESMTP id LAA20595
	for <alsa-devel@jcu.cz>; Tue, 14 Jul 1998 11:59:29 +0200
Received: from algorythmics.mhsg.ac.at (ritsch@algorythmics [193.170.129.58])
	by iem.mhsg.ac.at (8.8.8/8.8.8/Debian/GNU) with ESMTP id LAA03209;
	Tue, 14 Jul 1998 11:55:05 +0200
Received: (from ritsch@localhost)
	by algorythmics.mhsg.ac.at (8.8.8/8.8.8) id LAA00387;
	Tue, 14 Jul 1998 11:57:28 +0200
Date: Tue, 14 Jul 1998 11:57:28 +0200
Message-Id: <199807140957.LAA00387@algorythmics.mhsg.ac.at>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
From: Winfried Ritsch <ritsch@iem.mhsg.ac.at>
To: alsa-devel@jcu.cz
Cc: "P.J.Leonard" <P.J.Leonard@bath.ac.uk>
Subject: Re: ALSA Sequencer
In-Reply-To: <Pine.LNX.3.96.980713115108.502C-100000@entry.jcu.cz>
References: <35A5CF46.7DE1@bath.ac.uk>
	<Pine.LNX.3.96.980713115108.502C-100000@entry.jcu.cz>
Reply-To: alsa-devel@alsa.jcu.cz
Sender: alsa-devel-owner@alsa.jcu.cz
Precedence: list

> 
 > Note: I have still trouble with MPU-401 UARTs and compatible (this
 > includes all SB16/AWE soundcards). I don't know how can solve problem
 > with polled transmit (this means that CPU must wait if some room is in
 > transmit FIFO and after this send MIDI byte - no transmit interrupts are
 > generated). Looks that SB AWE 64 have only room for 2 bytes in transmit
 > FIFO!!! This is very small value and ALSA driver is able send with current
 > implementation (system timer is used) only 200 bytes per second on
 > i386 and 2048 bytes on alpha.
 > 
 > The problem maybe solved with some high-rate timer, but seems that SB16
 > doesn't have any timer connected (problem for .
 > 

I hope I dont say something common but here my experience:

a) The point is the scheduling intervall, which is for pentium computers
very conservativ, since 10ms on a i386 30MHz works ok. 1 ms on a 100MHz
is fine and 0.5ms on a 130 MHz wworks ok. I always run
"pd" on my computer (a 100MHz Pentium Notebook) just with 1ms, so also
reaction time to soundprecessing is reduced to 30ms instaed of 200ms.

I think 1ms has the alpha, so with 0.5ms there is no loss anymore.

do in include/asm/param.h #define HZ 1000 (or 2000 instaed of 100)

and recompile the kernel. Anyway I hope in future kernels this would
become default. So if anybody need fast MIDI he/she has to recompile
the kernel with this modification.

b) make a rtl-device which drives the MIDI and/or the sequencer, so
for good musicreaction use the rtl-patch for the kernel
 

mfg winfried

--- DI Winfried Ritsch - ritsch@iem.mhsg.ac.at ---
 INSTITUT FUER ELEKTRONISCHE MUSIK-
 University of Music and Dramatic Art
 Tel. ++43-316-389-7210, Fax.++43-316-389-7008

