Path: utzoo!attcan!lsuc!maccs!cs4g6ag
From: cs4g6ag@maccs.dcss.mcmaster.ca (Stephen M. Dunn)
Newsgroups: comp.sys.ibm.pc
Subject: Re: 640K limit
Message-ID: <25B5F23B.28498@maccs.dcss.mcmaster.ca>
Date: 18 Jan 90 16:43:39 GMT
References: <4668.25aed7f2@uwovax.uwo.ca> <1468@blackbird.afit.af.mil> <28808@amdcad.AMD.COM> <729@jethro.Corp.Sun.COM> <4308@brazos.Rice.edu>
Reply-To: cs4g6ag@maccs.dcss.mcmaster.ca (Stephen M. Dunn)
Organization: McMaster University, Hamilton, Ontario
Lines: 54

In article <4308@brazos.Rice.edu> solomon@screech.rice.edu (Richard L. Solomon) writes:
$In article <729@jethro.Corp.Sun.COM> acm@sun.UUCP (Andrew MacRae) writes:
$>In article <28808@amdcad.AMD.COM> phil@pepsi.AMD.COM (Phil Ngai) writes:
$>>Don't be ridiculous. The 8088 can't ADDRESS more than 1 mega. Just where
$>>did you expect IBM to put the IO stuff?
$>Simple.  IBM/MicroSoft *should* have used soft pointers to the I/O memory
$>areas.
$	NO....they SHOULD have mapped the I/O in the I/O ADDRESS SPACE where
$it belongs......there are 64kB of I/O addresses, anyone use that much?
$(Excluding VidRAM and EMS which obviously could legitimately be in the
$memory address space.)  

   Actually, have a look at the IBM PC memory map.  Up to segment A000 is
reserved for RAM.  A000-AFFF was reserved from day one for more capable
graphics cards (and has been used since EGA).  B000-BFFF is reserved for
display adapters (initially, B000-B0FF for the monochrome card, and B800-
BBFF for the CGA, with the rest to be used later - and it has been).
C000-F3FF is reserved for BIOS extensions - as in ROMs that belong in the
address space (this has also been used for RAM buffers on some cards,
such as IBM's Token Ring adapter).  F400-FFFF is for system ROM, IBM's
ROM BASIC, and the ROM BIOS.  The only ways to cut this down would be to
decrease the amount of space available for BIOS extensions (yeah, this
could be done, at the cost of more frequent collisions between cards)
or to have left less than 128K for display adapters and enforced the
sort of scheme used on VGA cards.  Oh, back to the token ring type of
card, which has its own RAM (up to 64K) on board:  if the I/O space
were to be reduced, it would have to use conventional memory for its
buffers.  Net gain for networks (which are becoming more and more
prevalent):  zero.

   So to get more memory, we would have had more awkward video cards and
more frequent problems when installing a new card in the system.  Worth
it?  Perhaps.  Don't get me wrong - I'm not trying to be a staunch IBM
defender (I don't think they came up with the best PC they could have, and
I don't think Microsoft came up with the best OS they could have, either);
I'm just pointing out that there isn't as much slack in there as one
might think there is.

$	Also, most early CP/M systems had 16 or 32kB of memory....their
$apps were limited to correspondingly less memory  :)

   Most early MS-DOS systems had 16 or 64K of memory ... perhaps IBM
didn't see their little computer growing so quickly that 10-40 times
its current memory would no longer be enough?  Or perhaps they were
under the impression that there would be an easier path to expansion
than Intel provided, so that users didn't have to recompile their
applications in order for them to take advantage of more advanced
processors (although I'm not sure how this would happen, exactly,
given how the limitations of the 8086/8088 come to pass).
-- 
Stephen M. Dunn                               cs4g6ag@maccs.dcss.mcmaster.ca
          <std_disclaimer.h> = "\nI'm only an undergraduate!!!\n";
****************************************************************************
       "I want to look at life - In the available light" - Neil Peart