From: mccalpin@gmp246.austin.ibm.com (McCalpin)
Newsgroups: comp.arch
Subject: Re: How does a chip get designed?
Date: Mon, 4 Mar 2002 17:06:46 +0000 (UTC)
Message-ID: <a609j6$dvi$1@ausnews.austin.ibm.com>

In article <a600al$m2i$2@news.rchland.ibm.com>,
Del Cecchi <cecchi@us.ibm.com> wrote:
>In article <a5vdrp$nqq$1@usenet.Stanford.EDU>,
> engler@Stanford.EDU (Dawson R Engler) writes:
>|> In article <a5tuis$fj4$1@pegasus.csx.cam.ac.uk>,
>|> Nick Maclaren <nmm1@cus.cam.ac.uk> wrote:
>
>Snip
>|>
>|> If by the above you mean that it's difficult to work on
>}> the boundary of two fields, then I think you will have
>|> to qualify this judgement somewhat before it becomes true.
>
>I believe the original comment had to do with being in a
>Oceanography department when one's interests and research
>had drifted towards HPC.

Almost....

I was working in the field of numerical ocean modelling,
which consists of writing programs that calculate approximate
solutions to various sets of partial differential equations
that, in turn, approximate the reality governing fluid flow.

I thought it was reasonable to investigate questions of the
class: "Is my model giving results that are accurate
approximations to the mathematical system that I intend
to be solving?".

I was wrong -- the promotion and tenure committee of my
college went so far as to say that 1/2 of my publications
"did not count", because they were either "applied math"
or "computer science", not "oceanography".  The unspoken
rule that I learned at significant cost was that it is OK
to spend your life writing and running numerical models,
but it is not OK to probe too deeply into questions of
mathematical correctness.

I am certainly not the first person run out of the field
of ocean modelling for daring to address the issue of
correctness, and it is likely that others will make the
same mistake I did.  Of course, it is possible to address
these issues from outside the field -- some excellent work
has been done by various mathematicians over the last decade
(notably Gerald Browning) -- but it is very hard to get such
work published in the traditional physical oceanography
journals.

--
John D. McCalpin, Ph.D.           mccalpin@austin.ibm.com
Senior Technical Staff Member     IBM POWER Microprocessor Development
    "I am willing to make mistakes as long as
     someone else is willing to learn from them."