Re: Microsoft edging toward the brink?

["Bryan J. Smith" <b.j.smith@ieee.org>]

On Sat, 2005-02-12 at 09:27 -0600, Robert G. (Doc) Savage wrote:
> Here's an interesting bit of fantasy or forecasting (take your pick)
> from someone who could probably care less about Linux.
> http://abcnews.go.com/Business/print?id=88655

Actually, I think he's got it exactly correct.  My favorites ...

  "Microsoft has always had trouble with stand-alone applications,
   but in its core business it has been as relentless as the Borg."

  "[Microsoft] is still a well-run company, which argues that its fade
   will be long and slow, like DEC, rather than a sudden death like
   Wang."

  "but, like Yahoo! and Apple before it, Microsoft may have to die in
   order to be reborn."

Microsoft isn't about making consumer products, it's about control of
the distribution channel.  It's that control that makes it very tough to
stop.

But in reality, Microsoft has lost _total_control_ over its ability to
deliver something usable for companies.  Now that they've gotten serious
about security, _everything_ breaks.  Even some of its own enterprise
apps won't run under XP SP2, just like NT 4.0 and 2000 before it (i.e.,
they "unhacked" things added in XP from 2000).

This all goes back to "Chicago" and Gates' decision to move forward with
DOS7 which became Windows 95.  Had Microsoft stuck with the original
Win32/NT model, the Windows world would be very different today.  But
because they didn't, and "Chicago" libraries and core services dominate
all versions of Windows today -- security is _impossible_.

And that's why Longhorn is _not_ the .NET model.  Unlike Win32 and NT,
Microsoft is not bothering with a pure .NET OS this time around, because
they know they will just break it anyway for compatibility.  Which is
why MS Office 2006 and countless other OS/apps are still 100% _legacy_
"Chicago" compatible.  And the Windows world looks pretty bleak.

Especially considering "Chicago" code is _stuck_ at x86 32-bit and
completely _unportable_.  That's where 32-bit/PAE36 virtualization atop
of 48-bit/PAE52 is going to save them because 64-bit versions of Windows
currently tend to be _slower_ because of WoW (Win32 on Win64) function
calls.

> I've been suspecting the same might be true of Intel in its on-going
> battles against AMD, where AMD very clearly has the upper hand in terms
> of products and technology vision.

It all depends on if Intel delivers "Yamhill2" like I've been
predicting.

Intel has done "Right Thing #1" -- _kill_ Pentium 4.  That bastard
needed to die, and it was a total waste of time -- especially all those
useless extensions that are designed for benchmarks, not mathematical
accuracy.  The LGA-775 platform has also been a major mistake.

The Pentium M aka "Pentium 3 revisited" is the stop-gap future for Intel
until "Yamhill2" makes it out the gate.  Until Yamhill2 does, if it
exists at all, Intel still hasn't fully designed a new 32-bit core in 12
years.  So they need to beat AMD to virtualization if the want any say.

AMD, on the other hand, suggested virtualization the moment the x86-64
manuals became available in 2002.  So you can be sure their next
generation processor in coming in the next year.  The Athlon design
(scaling from 500MHz to almost 3GHz) is 6 years old, just like its
NexGen predecessor had lasted 6 years before that (scaling from 84MHz to
over 500MHz).  We'll have to see what AMD has cooked up.

The future is virtualization, dozens of units, massively reordered and
renamed instructions and registers, etc...  Traditionally this will look
like multiple cores and threads of legacy ISA (instruction set
architectures) to traditional OSes.  To advanced OSes like Linux, this
will look like a massively parallel, reordering CPU.

Which makes me wonder what Cell is up to.


-- 
Bryan J. Smith                                  b.j.smith@ieee.org 
------------------------------------------------------------------ 
Community software is all about choice, choice of technology.
Unfortunately, too many Linux advocates port over the so-called
"choice" from the commercial software world, brand name marketing.
The result is false assumptions, failure to focus on the real
technical similarities of implementations and blind alignments.



-
To unsubscribe, send email to majordomo@silug.org with
"unsubscribe silug-discuss" in the body.