| Tuesday, March
20, 2001 Well I was wrong
about one thing in
my column: some of the AMD developed boards which I thought were not
out yet apparently are and have been "since Christmas" said one correspondent;
meaning I presume since Mid-January at latest. That column was written on
March 1, so I am about 6 weeks behind and that's not good. I have given
the relevant information in mail. It
also shows that one should not respond to reader pressure. What happened
was that the February column had a short bit intended to make people
reflect on benchmarks, and using the old Wintach as an example. It didn't
mention AMD at all, because for reasons not clear to me Wintach wouldn't
run on the Athlon system. It does now, which is a bit of a concern, since
the only change was turning the system off and turning it back on again.
When software won't run on a machine that has been running for months
without reset, then will run after a cold start, you want to know why. I
don't know why, but I have had mysterious happenings like that with VIA
chipset board (regardless of processor) just often enough to be worrisome
without being frequent enough to get statistics. Bill Godbout taught me
that that if an error rate was high enough to measure it's too high,
and that's still a good principle. The VIA error rate hasn't been quite
high enough to measure, but it was high enough to notice. Which could mean
it is coincidence. That is the major problem about not
being the old BYTE. In the old days I'd have turned this over to the test
labs and the geniuses in Peterborough. I don't have those resources now. So.
I didn't say anything about AMD and VIA in the February column. Whereupon
I got a ton of email demanding to know why. I should have realized that
this was a concerted effort and that most of the people sending that mail
had not even READ the column and didn't intend to read it; there is, apparently,
a fairly well orchestrated effort out there to influence journalists. One
is reminded of Guy Kawasaki's guerrilla journalism for Apple, which
resulted in many columnists simply refusing to mention Apple ever because
of the floods of hate mail that would result from anything less than
fulsome praise. But like an idiot I actually attempted to
pull together what I did know, which turned out to be about a month out of
date, and put together some reasoned observations about AMD and Intel. My
conclusion was that it doesn't matter a lot unless you are in specialized
applications in which case you will know a lot more than I do and don't
need my advice; and that the conservative (read failsafe) position is
avoid VIA until their next generation comes out, and the ultra conservative
position is stay with Intel because they have the resources, the track
record, and the incentive to make things work properly. I think I made it
pretty clear that if you like to experiment and you have the time, AMD is
certainly faster. Perhaps I did not make it clear that I was talking
mostly to those who want Good Enough Now and don't care about a bit more
performance for a hundred dollars or so less at the risk of having to put
more time into it. What you may expect by going AMD rather
than Intel is given by a reader in
an interesting comment. I don't have any reason
to change that conclusion. Intel is the safe way to go if what you want is
to have "good enough" without concerns. If you are out for the
latest and greatest at the lowest cost one can reach different
conclusions. I do have good reason to look into the latest -- and by that
I mean "since Christmas"
developments in this, and I will. My
thanks to those who took the trouble to point out where I was wrong
without sending nastygrams. My minor apologies to most nastygram writers
for not reading the long attachments which may have contained useful
information, but which I just didn't feel like wading through after the
opening (or often the heading). And to myself, a fooeygram: I should not
have let a concerted campaign of people accusing me of ignoring AMD push
me into precipitous action; which means, to the
readers an apology since I did write a column
segment based on information about 6 weeks out of date. And
I am still dancing as fast as I can. For as much as you can possibly want
to know, go to MAIL and follow the
threads. Finally, regarding Allchin
and the GPL discussion my choice of ADA was a poor example. It also
wasn't intended to be taken seriously: I was looking for an absurd
extension to a principle. Apparently I didn't make that clear enough, and
some people thought I really meant that using GPL would put compiler
outputs into public domain. My point was that no one would accept that. Acquinas
teaches us that the only way to have a discussion is to find some common
ground, some principles and propositions on which we can all agree, and
try to go from there. Often the best way is to find what is absurd: if we
can agree that x is absurd we can assume not-x in future discussions. I
often do this in my essays, but this time I apparently didn't make it
clear enough that I thought everyone would agree on the absurdity of the
proposition that since ADA was developed with public money, any program
compiled with ADA would be public domain. My apologies. As
to Allchin's remarks I thought then and think now that they were a trial
balloon, and had they got any general acceptance they would not have been
"explained away" as they were in the Microsoft
"explanation." Having said that I see no reason not to accept
the Microsoft explanation and to put as good a face on it as possible. The
explanation actually concedes quite a lot, and I think that all to the
good, and I see no point whatever in dwelling on "what he really
mean". The outcome of all this is that Microsoft has taken an
official position that doesn't look all that bad for GPL. Let's take it as
common ground for the next discussion. One should always
build golden bridges for one's enemy's retreat... And
Bob Thompson says I may not have been as remiss as I was persuaded. See
mail. Which makes me feel much better. Better to make an apology when
it is not needed than to need to and not do it. And the
market is down again. And again. I could have sold off some funds (for
obvious reasons I only invest in funds, not in individual stocks) and paid
the mortgage... Ah well. Timing is everything. And "of
all sad words of tongue and pen, the saddest are these, 'it might have
been.'" On the subject of compensation for effort, go
to http://www.irfanview.com/english.htm
and download the program, then wrap up a ten dollar bill and mail it to:
Irfan Skiljan
Postfach 6
2752 Woellersdorf
Austria, Europe
The program is freeware, but it's worth supporting
him. He has no way to cash checks or take credit cards. so there's no
choice but to send cash. Do it before you forget.
TOP |
CLICK HERE for
a Special Request. 
ALERT