From mboxrd@z Thu Jan  1 00:00:00 1970
Received: (from weis@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id KAA20964 for caml-red; Wed, 18 Oct 2000 10:39:05 +0200 (MET DST)
Received: from concorde.inria.fr (concorde.inria.fr [192.93.2.39]) by pauillac.inria.fr (8.7.6/8.7.3) with ESMTP id DAA14762 for <caml-list@pauillac.inria.fr>; Wed, 18 Oct 2000 03:17:22 +0200 (MET DST)
Received: from unknown-147.101.pilot.net (unknown-147-101.pilot.net [198.232.147.101])
	by concorde.inria.fr (8.11.1/8.10.0) with ESMTP id e9I1HKX14681
	for <caml-list@inria.fr>; Wed, 18 Oct 2000 03:17:21 +0200 (MET DST)
Received: from unknown-24-15.pilot.net (unknown-24-15.pilot.net [206.189.24.15]) by unknown-147.101.pilot.net with ESMTP id SAA06299 for <caml-list@inria.fr>; Tue, 17 Oct 2000 18:17:19 -0700 (PDT)
Received: from uswaumsxbhmedge.med.ge.com (localhost [127.0.0.1]) by unknown-24-15.pilot.net with ESMTP id SAA23035 for <caml-list@inria.fr>; Tue, 17 Oct 2000 18:17:18 -0700 (PDT)
Received: from mrs.mrs.med.ge.com (3.57.196.1 [3.57.196.1]) by uswaumsxbhmedge.med.ge.com with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.2651.58)
	id 49ZYW494; Tue, 17 Oct 2000 20:17:17 -0500
Received: (from gurr@localhost)
	by mrs.mrs.med.ge.com (8.9.3+Sun/8.9.3) id SAA07872;
	Tue, 17 Oct 2000 18:17:07 -0700 (PDT)
Date: Tue, 17 Oct 2000 18:17:07 -0700 (PDT)
From: David Gurr <gurr@mrs.med.ge.com>
Message-Id: <200010180117.SAA07872@mrs.mrs.med.ge.com>
To: caml-list@inria.fr
Subject: Re: single-precision floats, etc.
Cc: gurr@mrs.mrs.med.ge.com
X-Sun-Charset: US-ASCII
Sender: weis@pauillac.inria.fr

Chris Hecker:
> Most importantly, is there any way to make the "float" type compile to single precision floating point numbers?  

I do mostly DSP stuff for MRI machines and most of the code is single
precision floats.  Instead of adding single precision floats, I
wonder if I would rather replace boxed float with "boxed 
multi-media datatype".  So in Proc and Arch for AltiVec machines the
float registers would be replaced by 128 bit "vector" registers
and float arrays would instead be arrays of 128 bit elements and the
float ops would be replaced by packed int and packed float ops.

On the other hand, most x86 machines have weird multimedia
instruction sets (using the MMX registers precludes using the 
floating point stack, clever).

I suppose that Cmm would be modified to either replace the
float ops with packed ops or extend the float ops with 
packed ops and Selection would do what it could with 
incomplete multimedia instruction sets.

-David Gurr