From: "Tony O'Bryan" <aho450s AT nic DOT smsu DOT edu>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: MMX
Date: Tue, 26 May 1998 18:35:45 +0000
Organization: Southwest Missouri State Universtiy
Lines: 19
Message-ID: <356B0B80.974EE5EF@nic.smsu.edu>
References: <356A673C DOT FEB310D2 AT mailexcite DOT com>
	 <MPG DOT fd4fd49e501ce63989684 AT news DOT virgin DOT net> <6NFZJBA16wa1Ew7Y AT talula DOT demon DOT co DOT uk>
NNTP-Posting-Host: marie.a48.smsu.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

Shawn Hargreaves wrote:

> Ken writes:
> >I think you could still use the MMX instructions, as I understand it the
> >instructions enable MMX machines to process them, and also machines
> >without MMX can still process them.
>
> Not true. If the machine doesn't have MMX, these instructions cannot be
> used. You can't really expect something to work even when the hardware
> doesn't support it...

The '386+ chips have an "unknown op-code" exception that can be intercepted and
redirected to an instruction emulation routine in software.  Granted, there is
no practical reason to try emulating the MMX instructions since the performance
drain will render the MMX-enabled software next to useless anyway.

But it is possible.