Date:	Fri, 6 Jun 1997 09:14:05 -0400 (EDT)
From:	"Art S. Kagel" <kagel AT ns1 DOT bloomberg DOT com>
To:	Paul Derbyshire <ao950 AT FreeNet DOT Carleton DOT CA>
Cc:	djgpp AT delorie DOT com
Subject:	Re: FreeWin95 Project
In-Reply-To:	<5n32mq$sj1@freenet-news.carleton.ca>
Message-Id:	<Pine.D-G.3.91.970606090636.12912A-100000@dg1>
Mime-Version:	1.0

On 4 Jun 1997, Paul Derbyshire wrote:

> 
> Josef Moellers (mollers DOT pad AT sni DOT de) writes:
> > I beg to differ.
> > It's the processor hardware that determines the endianness. The OS has
> > to live with it or ... die.
> 
> As long as the CPU isn't bytesexual, yep.
> 
> > There are CPUs that can be both, e.g. the MIPS CPUs can switch between
> > big endian and little endian mode.
> 
> Those are called "bytesexual". There're a few like that. Not all Unices
> run on bytesexual machines. (In fact a unix of some kind or other can be
Actually I've been told be people who should know that the Pentium and 
Pentium Pros have an Endian switch just like the MIPS processors, though 
I do not know of anyone who is using it.  Indeed at Bloomberg we have 
much BigEnd dependent code (legacy from old Perkin-Elmer CPUs which 
survived well on Data General's M88110 Aviion systems).  We must now move 
to an Intel platform and DG has produced compilers and OS drivers to 
support our code WITHOUT switching the CPU to BigEndian mode as they are 
not willing to port DGUX/Intel to the Pentium in BigEndian mode.  So it 
IS possible to run BigEndian on a Little Endian machine.  So far the only 
performance problems not yet licked involve floating point data and DG 
has just started working on that.

> > No, it is not directly possible to "override" the endianness of the CPU.
> 
> Unless it's a bytesexual CPU.
> 

Art S. Kagel, kagel AT bloomberg DOT com

- Raw text -