Newsgroups: comp.os.minix
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!m.cs.uiuc.edu!ux1.cso.uiuc.edu!bradley.bradley.edu!buhub!drp
From: drp@buhub.bradley.edu (Douglas Pokorny)
Subject: Re: 386sx/dx
Message-ID: <1991May2.171330.27131@bradley.bradley.edu>
Sender: news@bradley.bradley.edu
Nntp-Posting-Host: buhub.bradley.edu
Organization: Bradley University
References: <52442@nigel.ee.udel.edu>
Date: Thu, 2 May 91 17:13:30 GMT
Lines: 40

nicolas@cli52or.edf.fr writes:

>16-bit bus interface : 
>does this means that the segment's size of a process is limited to 64K ?

Nope, externally the 386sx has a 24-bit address bus and a 16-bit data bus.
Internally the 386sx has 32-bit registers and the same 64-terrabyte
virtual memory management hardware as the 386dx.

As previously mentioned, the only differance between the 80386dx and the
80386sx are:

Physical packaging/size (sx is a gull-wing flat-pack, dx is a PGA device)
Size of the the external addressing lines --
16-bit data -vs- 32-bit data and 24-bit address -vs- 32-bit address.
Size of the instruction pre-fetch queue.

There technically isn't ANY software which will run on a 386dx that won't
run on a 386sx.  I just wish that computer salesmen and the idiots in the
world would get that through their heads.

The only software which will operate differently is one which tries to
check the pre-fetch queue size.  Source code for this is on Simtel20.
This however is NOT normal "run-of-the-mill" programming, and anyone
who would use the above to limit the processors on which their software
will run has got to have a serious mental deficiency.

-Douglas


_________________________________________________________
 ________   ___    ___  
|_   __  \ |   |  |   |   Douglas R. Pokorny
  | |__| /  | |    | |    drp@buhub.bradley.edu
  |  __  \  | |    | |    CS major/Geisert Hall Resident
 _| |__|  | |  \__/  |    The best of both worlds!
|________/   \______/     
  Bradley   University    "The Harvard of the Midwest"