Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!att!cbnewsl!bonnie!rbr
From: rbr@bonnie.ATT.COM (4197,ATTT)
Newsgroups: comp.sys.ibm.pc.misc
Subject: Re: When will the 8088 die?
Message-ID: <1990Dec13.124558.5832@cbnewsl.att.com>
Date: 13 Dec 90 12:45:58 GMT
References: <90335.202651F0O@psuvm.psu.edu> <3360005@hpsgwp.sgp.hp.com> <1990Dec4.014539.13773@magnus.ircc.ohio-state.edu> <1990Dec4.160730.15617@alchemy.chem.utoronto.ca> <1990Dec5.152318.7316@cbnewsl.att.com> <59765@microsoft.UUCP>
Sender: @cbnewsl.att.com
Reply-To: rbr@bonnie.ATT.COM (Bob Rager)
Distribution: na
Organization: AT&T Bell Laboratories
Lines: 49

In article <59765@microsoft.UUCP> steveha@microsoft.UUCP (Steve Hastings) writes:
>In article <1990Dec5.152318.7316@cbnewsl.att.com> rbr@bonnie.ATT.COM (Bob Rager) writes:
>>But who cares, isn't MessyDOS the real culprit that determines this
>>stupid architecture.  Is it not M*DOS that puts the BIOS and video RAM
>>at the "top" of memory?
>
>Umm... no, it is not MS-DOS that arranges the hardware in a computer.  IBM
>did it, when they designed the PC, and they had seemingly good reasons at
>the time.  When they first made the IBM PC, 16k was standard and 64k was a
>lot of memory.  640k seemed enormous.  You can't blame them for this.

I can remember mainframes when 128K was a BIG machine.

As an X mainframe guru for the now desolved Western Electric and an
IBM watcher of many years, I most heartily agree. I appreciate IBM's position,
when you are the first you have a great chance of making the wrong coice.
Then we all have to live with it forever. 

The PC came out when several machines had 128K (Commodore 128, TRS-80 IV),
as I remember the original PC had 128K and a cassette port.  I don't know
who might be at fault but it would have been so easy to put the system stuff
between 0 and nK and provide a pointer to the base of user memory.  That
way we could have a linear memory space all the way up to 3+ GB. Then it
would only be a matter of money.  BTW, I have heard that the IBM chose the
group designing a PC using the 8088 because they thought it would be the 
PC most likely to fail in the marketplace as they were afraid that PC's
would eat into their Mainframe sales ($$$$$$).

>
>You can safely flame IBM for the design of the BIOS, though.  The BIOS
>functions were too slow, so everyone wrote straight to the hardware.  Thus
>if you want to run MS-DOS programs you need an exact copy of the PC
>hardware, or a good emulation thereof.  In some strange, alternate universe
>where IBM's BIOS was written better and everyone used it, all you need to
>run DOS programs is a faithful emulation of the BIOS and the CPU.

If you would dis-assemble the BIOS you would find that you would have 
a hard time making it more compact.  The efficiency  problem arises when
you try to write completely generalized routines.  The CYA code becomes 
several times the size of the do it code.  The trick might be to execute 
BIOS from super fast RAM.
 
>-- 
>Steve "I don't speak for Microsoft" Hastings    ===^=== :::::
>uunet!microsoft!steveha  steveha@microsoft.uucp    ` \\==|

Lets not have a flame war over this. 

Bob Rager