Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!rpi!raster
From: raster@itsgw.rpi.edu (Jerry D Bain)
Newsgroups: comp.windows.ms
Subject: Re: IBM's XGA and Windows
Message-ID: <~'0^KN@rpi.edu>
Date: 28 Jan 91 14:05:51 GMT
References: <859@nih-csl.nih.gov> <1991Jan24.132327.8830@gpu.utcs.utoronto.ca> <4126@ns-mx.uiowa.edu> <1991Jan24.201315.647@sbcs.sunysb.edu> <1991Jan28.100703.4680@cc.helsinki.fi>
Lines: 21
Nntp-Posting-Host: jec407.its.rpi.edu

tmottonen@cc.helsinki.fi writes:

>    I've heard that the XGA adapter requires a 386SX or better
>    machine. Any idea why ? 

The XGA's memory-mapped color-maps are typically placed in really high memory
areas... like above the 3-4 gigabyte address range to prevent any software
memory conflicts in the near future (unless someone gets the bright idea to
write an application that actually needs 4 gigabytes on a '386).

This frees up the first meg for normal DOS stuff...  This means that in *XGA
mode* only, the VGA memory area is freed up for DOS applications.  In VGA
mode, everthing works as before.  In addition, up to eight XGA cards are
supported in a machine (since each can have a different address).

In addition, most of the actual software drivers require access to the 
priveledged '386 instruction set.  At first blush, this '386 requirement may
seen heavyhanded by IBM (the only manufacturer of a XGA board to date), but
all in all it adds up to the first *well thought out* display standard
for the iAPX86 architectures I have seen to date.  It will last a long time,
hopefully.