Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!pasteur!ucbvax!decwrl!labrea!rutgers!att!occrsh!rjd
From: rjd@occrsh.ATT.COM (Randy_Davis)
Newsgroups: comp.misc
Subject: Re: Ethics of crippler circuitry
Summary: Do it, but be aware and accept the consequences - 3B2 example
Keywords: 3B2, clock, speed, manufacturer intentions
Message-ID: <570@occrsh.ATT.COM>
Date: 14 Feb 89 17:53:43 GMT
References: <7143@pyr.gatech.EDU> <11630010@hpsmtc1.HP.COM> <1257@raspail.UUCP> <132@aucis.UUCP>
Reply-To: rjd@occrsh.UUCP (Randy_Davis)
Organization: AT&T Network & Data Systems, OKC
Lines: 93

In article <132@aucis.UUCP> bnick@aucis.UUCP (Bill Nickless) writes:
|What do the companies (Amdahl, IBM, et. al.) do about the "sneaky" computer
|operator who figures out that you remove board X and get Y*1.5 performance?
|Or if you flip this switch or comment out this code or patch the OS to allow
|more than the licensed number of users on?  

[  The following is the relating of a *personal* incident and in no way     ]
[ am I an official representative of AT&T or their policies.  The following ]
[ is completely personal opinion.]

  On the 3B2 mailing list last year, one of the more verbose contributors wrote
a quick message saying that, if you had a brass plate on top of your processor
chip set on the system board, you could just switch out the clock with a faster
one and have a faster machine.

  As I work on 3B2's and supervise portions of the testing at the factory, I
was aware that the brass plates were not related to the rated clock speed of the
system boards and said as much (the maximum rated speed of the processor chips
is set through actual testing of the chips in question, the brass plated ones
were just different production runs for other reasons).  I went ahead to say
that it was *possible* that the machine would work at the higher speed (the
speeds in question were 10MHz stock vs the 14MHz "upgrade"), yet a complete
suite of tests would be necessary to have any confidence in this.  I also went
on to mention that it is not something you do without testing it completely
first.  All I got was written abuse for taking the time to warn him (when my
sole purpose was to prevent people from taking his uninformed word and to warn
him and others of the possible problems).

  My point *is* valid too, as, since the board in question is one of the ones
under my supervision, I have seen many types of strange hardware-caused errors
to do with system speed.  Many of which it is possible might not manifest
themself unless the user does a certain combination of commands (I recall one
board who liked to dump core on *one* certain Unix System V operating system
command which turned out to be caused by a bad MMU).  It was this danger that
other 3B2-owners might take his erroneous advice and end up with screwed up
machines that I was worried about.  It is also probably the reason that the
upgrade kit for the machine contains a new processor chip set rated for the
higher speed.

  It is really funny in some ways:  since the email and mailing list war ended,
the person in question posted a few remarks about his "upgraded" system
and its (new and higher) benchmark numbers, etc...  One of which was ended
with the line "I am posting this just to make a point" that I can only assume
was aimed at me.

  Hilarious!!!  As if a faster clocked machine is *not* going to post better
benchmark numbers???  I was not concerned with its bench-marks (I knew they
had better be higher if it worked at all), just with its complete functionality.

  I was personally very happy that it *did* work at the faster speed, as
perhaps it is proof that the team I am on is doing a good job of stress testing
the machine before they leave the factory.  But, sure enough as his *does*
work at the higher than intended speed, probably the next 100 won't.

  All this uninformed "upgrading" from a (claimed) electrical engineer who
should know better, or should at least admit the fact that you should not
operate a piece of equipment outside its specifications without knowing WHY
the specifications were set or how to test the new COMPLETE functionality of
the machine.

|How does that concept of "I've bought it, it's mine" apply here?  Is it 
|unethical to get the best performance possible out of a machine you've paid
|for, even if the manufacturer didn't intend for that performance to be avail-
|able?
|
|If a vendor is going to sell me a computer, I would use it to its best 
|performance--regardless of what the vendor intends!

  I personally really do believe that Bill is correct here.  If you bought it,
you should be able to do what you want to it. 

  ***BUT**!!!!  Do not assume that the manufacturer is unaware of the possible
"upgrades" you might attempt.  First, you might ask yourself (or others that
might be in the know) for what reason the manufacturer might want to hold 
the higher ability of the machine away from the customer.

  One way to do this, as has been cited before, is to investigate what the
*manufacturer* does to upgrade the machine.  If, as in the above-cited case,
the upgrade *does* involve a hardware limitation, PROCEED WITH CAUTION, not
with the blind enthusiasm of a child with a new toy (as my 14MHz-happy friend
was sounding).   

 If the upgrade is simply switching a switch or trading out a known lower
spec part that is holding everything else down, *AND* you are sure that
that is all that is needed and that that is the way the manufacturer does
it, I would personally say DO IT!!  And DO IT the cheapest way you can....

  The whole problem is that the uninformed often think that the BIG BAD
companies are just wanting their money.  Well, sometimes it may be true,
but it is a dangerous generalization to make all the time.

Randy Davis					UUCP: ...(att!)ocrjd!randy
						      ...(att!)occrsh!rjd