There have been some interesting postings on this list regarding the
Compaq/Intel agreement. I'm sure there are many different opinions about the
agreement, but I would like to share mine.
Back in 1993, I attended UNIX Expo in New York City. It was like a college
reunion. The UNIX community was a small close knit group back then, and it
seemed like I recognized everyone there. The DEC booth was showcasing a DEC
3000 with its new Alpha chip running OSF/1. The performance was
enthralling... I was so impressed, I joined DEC soon after, working on
OSF/1.
Since then the Alpha chip has retained the performance lead. It's quite an
achievement for one chip to maintain a lead for so many years.
But Compaq had to naturally ask "will Alpha retain its lead, and for how
long?". An exhaustive study was done by Engineering (not the bean counters),
and it was determined that by 2004, Alpha's performance would not be
dramatically different from the Itanium chip from Intel. (Note that unlike
other companies, Compaq did this study several years before our chip ran out
of steam).
So Compaq said to Intel "we have the chip expertise, the compiler wizards...
we've been doing 64 bit for more than 10 years now... let's partner and make
the Itanium a performance leader".
Why Intel? Note that Intel has been making the Alpha chip since DEC settled
with Intel for Alpha patent infringements in the Pentium III. The
Alpha plant in Hudson, MA was bought by Intel in the terms of that
settlement.
(Samsung also produces Alpha in its foundries).
Thus it was a natural decision for Compaq and Intel to partner. The result
will be an "Alpha-bitized" Itanium that will incorporate the best from both
Alpha and Itanium.
For the customer, the term I've heard is that this is an "engine swap". I
don't know all the details yet, but binary compatibility is a goal, with
source code recompile the minimum acceptable migration path.
The result? Tru64UNIX will be on a more mainstream chip (Alpha was always
considered a beautiful design, but still not mainstream). Software
developers will port to Tru64UNIX/Itanium at the same time they do their
Solaris/ancient Sparc and AIX/PowerIV ports, and thus application
availability (which IMHO was the reason Alpha suffered) will no longer be an
issue. And since, IHMO, Compaq provides the best OS (Tru64UNIX), the best
clustering (TruCluster), the most reliable systems, everything should turn
out fine.
There are a lot of emotions tied up in this announcement, and a lot of soul
searching is going on within the Tru64UNIX group. I don't think it is too
presumptuous for me to say that we all appreciate your support of Tru64UNIX
both now and in the future.
Regards,
Paul
--
Paul Henderson
Tru64 UNIX Engineering, Compaq Computer Corporation
mail: Paul.G.Henderson_at_compaq.com || 200 Rt 9, Manalapan, NJ 07726
Received on Wed Jun 27 2001 - 18:04:30 NZST