From alsa-devel-owner@alsa.jcu.cz  Thu Jan 21 13:16:18 1999
Received: from gw-nl3.philips.com (gw-nl3.philips.com [192.68.44.35])
	by marvin.jcu.cz (8.9.1a/8.9.1) with ESMTP id NAA21003
	for <alsa-devel@alsa.jcu.cz>; Thu, 21 Jan 1999 13:14:15 +0100
Received: from smtprelay-nl1.philips.com (localhost.philips.com [127.0.0.1])
          by gw-nl3.philips.com with ESMTP id NAA28217
          for <alsa-devel@alsa.jcu.cz>; Thu, 21 Jan 1999 13:14:06 +0100 (MET)
          (envelope-from guenter.geiger@psp-vie.be.philips.com)
Received: from smtprelay-eur1.philips.com(130.139.36.3) by gw-nl3.philips.com via mwrap (4.0a)
	id xma028214; Thu, 21 Jan 99 13:14:06 +0100
Received: from sunnm.austria.cp.philips.com (sunnm.at.cp.philips.com [130.146.10.240]) 
	by smtprelay-nl1.philips.com (8.8.5/8.6.10-1.2.2m-970826) with ESMTP id NAA08077
	for <alsa-devel@alsa.jcu.cz>; Thu, 21 Jan 1999 13:14:03 +0100 (MET)
Received: from pspvie03.pspvie (pspvie03.psp-vie.be.philips.com [130.146.61.3])
	by sunnm.austria.cp.philips.com (8.8.8/8.8.8) with ESMTP id NAA15716
	for <alsa-devel@alsa.jcu.cz>; Thu, 21 Jan 1999 13:15:40 +0100
Received: from PC468 (pc-61-95.at.ds.philips.com [130.146.61.95]) by pspvie03.pspvie with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.1960.3)
	id CXKA2ZG9; Thu, 21 Jan 1999 13:14:47 +0100
From: "Guenter Geiger" <guenter.geiger@psp-vie.be.philips.com>
To: <alsa-devel@alsa.jcu.cz>
Subject: RE: hi-res pcm sound support/IEEE floats
Date: Thu, 21 Jan 1999 13:13:59 -0800
Message-ID: <000001be4582$f7406490$5f3d9282@PC468>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
In-Reply-To: <Pine.LNX.3.96.990121100624.28951C-100000@entry.jcu.cz>
Reply-To: alsa-devel@alsa.jcu.cz
Sender: alsa-devel-owner@alsa.jcu.cz
Precedence: list



> -----Original Message-----
> From: alsa-devel-owner@alsa.jcu.cz

> If a DSP for example expects data in 24-bit format (three bytes), why
> a conversion from float -> 24-bit should be done in the driver?
>

Im not sure if this conversion is even doable. Is it possible to use
floating point arithmetic in the kernel ?
Im not a specialist in hardware, but I have the feeling that it is rather
complicated to deal with 24 bit entities when transferring data to the
soundcard.
 Therefore I think we would rather see cards which exspect that
data is in 32 bit format.

> OK. We can speak how will things work inside alsa-lib. Here maybe all
> samples processed using float format and they should be
> converted to right
> (hardware) format finally for the driver.
>

Actually this is very good thing, and it would make the driver look even better
fit for professional work.


And to throw in some other thoughts ...
Motorala DSP chips are (mainly) designed to do fixed point arithmetic. They are
a lot faster with that, event is most likely using fixed point.
MMX is an extension which uses the FPU registers, but does integer calculations,
and *not* floating point.

despite of that, I totally agree on the floating point issue.

Guenter




