From:	Hans-Bernhard Broeker <broeker AT physik DOT rwth-aachen DOT de>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: mmx and gcc
Date:	4 Oct 2001 12:08:39 GMT
Organization:	Aachen University of Technology (RWTH)
Lines:	19
Message-ID:	<9phjg7$ktj$1@nets3.rz.RWTH-Aachen.DE>
References:	<9peu00$4rq$02$1 AT news DOT t-online DOT com>
NNTP-Posting-Host:	acp3bf.physik.rwth-aachen.de
X-Trace:	nets3.rz.RWTH-Aachen.DE 1002197319 21427 137.226.32.75 (4 Oct 2001 12:08:39 GMT)
X-Complaints-To:	abuse AT rwth-aachen DOT de
NNTP-Posting-Date:	4 Oct 2001 12:08:39 GMT
Originator:	broeker@
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

Benjamin Kalytta <bkausbk AT web DOT de> wrote:
> how can I use mmx in gcc ? 

GCC itself doesn't offer you any access to MMX at all.  The original
MMX is pretty useless for compiled code, anyway, due to the way it
interferes with usage of floating point numbers.  That may be
different for its later versions 3DNow! and SSE 1/2, to some extent.
But the further you go in this development, the more you limit the
range of platforms your code will work on.  

That leaves inline assembly on appropriately typed data as the only
way.  

You may easily end up in a code-forking nightmare if you're not very
careful.  And the fact that MMX and friends are rather hard to code
efficiently for truly doesn't help with that, either.
-- 
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.

- Raw text -