NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/alt/folklore/computer/16523

Text File | 1992-11-20 | 4.1 KB | 80 lines

Newsgroups: alt.folklore.computers
Path: sparky!uunet!mcsun!dxcern!vxcrna.cern.ch!zeus
From: [e-mail]
Subject: Re: ASSEMBLY LANGUAGE IBM
Message-ID: <[e-mail]>
Sender: [e-mail] (USENET News System)
Organization: CERN European Lab for Particle Physics
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Date: Fri, 20 Nov 1992 20:40:59 GMT
Lines: 68

In article <[e-mail]>, [e-mail] (Charles Lasner) writes:
> In article <[e-mail]> [e-mail] writes:
>>[e-mail] (Steve VanDevender) writes:
>>
>>>I suspect that the questions apply to IBM's most famous series of
>>>machines, the 360/370 series.  If you thought them placing
>>>increasingly souped-up versions of the 8088 in IBM PCs was bad,
>>>note that the 360/370 series has been using the same instruction
>>>set and processor architecture for nearly _30 years_.  IBM's
>>>top-end mainframes (like the 3090) still run IBM 360 programs in
>>>binary form.
>>
>>From this I infer that you consider it a ``bad thing'' that an IBM mainframe
>>(or PC) user can upgrade his hardware without buying new versions of commercial
>>software?
>
> All that was stated is that the binary object modules of older programs can
> be made to run on the new machine.  In practice, this may be inadequate
> depending on what the newer model actually has available on it.  Moreover,
> IBM has tended to proliferate excessive versions of their O/S on these
> machines, as if they actually were separate and incompatible.  Those of
> us used to either a relatively limited number of say VMS or unix variants
> don't appreciate the extraneous nature of artifically distinct O/S variants
> sold as separate entities.  This tends to make one rethink the worth of
> staying with a family of machines where you actually do have to keep buying
> what is substantially the same software again and again without the means
> to recycle it.
>
> In the early 370 days, IBM had a campaign to extraneously introduce 370-only
> instructions into major components of O/S, so that they couldn't run on
> 360's.  Various users distributed patches to remove the extraneous
> dependancies, so that purchased versions of O/S 370 could run on high-end
> 360's that were otherwise viable.  IBM wanted them plowed under and then
> make the customer buy the latest and the greatest.  But companies such
> as ITEL were refurbishing 360's and making them have new buss peripherals,
> so it was necessary to keep the software compatible in spite of IBM's
> attempts at planned obsolescence, and what effective amounts to restraint of
> trade in this situation (by making ITEL's product less attractive since it
> was claimed to only run obsolete versions of the software, etc.).
>
> cjl
The 370 extra instructions were mainly to kill Amdahl at birth.

The microcode of most of the 360 hardware meant that they could be
 upgraded with little hassle. The Amdahl machines were hard wired and
were thought to be unable to run the new O/S

IBM did most of the damage by annoncing the fact that new instructions
 would come out long before they actualy announced what they were.
When the instructions were finaly released Amdahl had them
copied in 6 months and retrofitted patches.

This set of games cost IBM $200 million when they were sued
by the European Community under anti-trust laws.

If IBM were not in a near bankrupt state anyway the Clinton
administration might revive the old
Carter anti trust suit which Regan stopped (IBM was big campaign contributor).

Hopefully HP and SGI will manage to grow enough to be a
counterweight to DEC. Once big blue goes under I wouldn't like to see
another monopoly situation develop. As it is DEC will have the
database and commercial sector sewn up unless HP and SGI manage
to make an entry on the back of Windows NT. Sun is trying the same
tactics DEC did in the 1980s - a marketing led company out to
screw the customer.