doubts

[jfox@xxxxxxxxxx (Jeff Fox)]

Dear MISC readers:

About all I can say about Mark's last post is:

Mark you keep bringing up the same irrelevant stuff over
and over.  Yes I know you have lots of doubts.  But you are
not listening to the answers.  You keep returning to issues
relating to a small class of grand challange problems and
saying these things are not addressed.

When we say we have a dozen benchmarks that reflect what the
machine is designed to do, you refuse to recognize them as
benchmarks and keep insisting that unless we show you something
irrelevant you have not seen a benchmark.

You keep ignoring the logic presented to you because it doesn't
fit your model, then you insult the person trying to help you
understand by and their argument by tring to classify it
as religion.

I went through this many times many years ago with many people.
"micros cant execute mainframe software therefore they will never
amount to anything."  and "I have never seen any evidence that
micros can be used for anything because I have not seen my
favorite mainframe benchmark yet running on this micro."

We have explained things, we have shown numbers, benchmarks,
spreadsheets, and given lots of explainations, but you keep
returning to things like "I haven't seen anything because I
have not seen SPECmarks."

I think the problem is not that you are being fed religious dogma
without logic, numbers, reasoning, benchmarks, examples etc, but
that the very idea doesn't match your religious ideas about
if it isn't designed for C, UNIX, and doesn't run SPECmarks it
isn't a "real" computer.

I also have little patience with people who repeatedly say
"price or price/performance is not important to me because price
is no object when tax dollarys pay for our government or university
computers."  You are not more important than everyone else.  And
you may learn that the people who are paying for such waste may
pull the plug someday.

I know improving price performance by orders of magnitude is of
no use if it also might require someone to think.  And after all
whether in corporate culture or academia the money spent (wasted)
in your territory is a measure of your status and sucess.

So in your case I would recommend staying with what you understand
and trying to get them to spend as much money as possible on it.
Don't waste your time also with new technology that might require
you to take time to rethink anything.  After all unless you take
the time to understand it you will have no hope of getting permission
to actually do anything with this kind of experimental stuff.  If
you did try to do something with this technology you might have
the problem that other people won't understand what you are doing.

I think you should stick with C and your workstation for whatever
it is you do besides run SPECmarks.

Jeff Fox