[KLUG Members] Re: Yo! -- _All_ CPUs had the "DOS timing issue" ...

[Bryan J. Smith]

--=-dj+xINrjE9lleBrubhQ/
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On Fri, 2002-11-29 at 18:05, Adam Bultman wrote:
> When I pressed him about it, he said that a lot of amd chips were
> even incompatible with windows itself!  (This was last year).
> When I told him that those were a very small range of chips
> (300-400 MHz K6-2s), and that a library change would fix it,

That was a _DOS_ "timing issue."  _All_ CPUs had it, even Intel's.  But
Microsoft included fixes for Intel chips in the release versions of
Windows 95, whereas they did not for AMD.

The "change in attitude" from Microsoft came when it realized that 5-10%
of its tier-1 OEMs and 30-50% of its tier-2 OEMs had sold K6 systems and
users were running into the issue.  Again, only when _volume_ matters
does Microsoft accomodate.

It had 0% to do with AMD.

I could go into the 4M "paging bug" of the Athlon was well.  Yet another
"Intel Optimization" that is _unnecessary_ on Athlon.  In fact, the
reason why Intel is now recommending people turn on the 4M paging option
is because the P6 core can_not_ match the Athlon core at paging with
today's memory sizes using only 4K.

Sad, really sad.  Chronic, "oh yeah, we didn't design X right in the
first place, but we offer Y instead."  Unfortunately, Y is not well
tested on _any_ chip (even Intel's).  And I could go into how AGP sucks
(and mega-unreliable in general) -- even Digital, who basically created
AGP (let alone PCI, which that's all AGP is, a glorified PCI bus) for
Intel, said so too (in a nutshell:  it was an Intel answer to not
putting more than one PCI bus in a system).

> he still held to his beliefs, despite the fact that clearly, AMD
> worked fine in all recent cases.  Ahhh....  He also didn't want me
> to run linux on my workstation.  I thought it was peculiar.

Again, as Linux users, AMD v. Intel is a non-issue from a
"compatibility" or even "performance" standpoint.  With AMD holding over
a 30% marketshare, and now selling a _lot_ of OEMs and developers on
x86-64, Intel's going to have trouble if it keeps reusing its 386/486
opcodes for SIMD.


--=20
Bryan J. Smith, E.I.            Contact Info:  http://thebs.org
A+/i-Net+/Linux+/Network+/Server+ CCNA CIWA CNA SCSA/SCWSE/SCNA
---------------------------------------------------------------
The more government chooses for you, the less freedom you have.

--=-dj+xINrjE9lleBrubhQ/
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQA95/iADjEszaVrzmQRAkUCAKCGz0YxXAImi09kP8dowehovaRF7wCgu2aP
2MfsbPRcHjGBNim1Yr4lREA=
=VdoF
-----END PGP SIGNATURE-----

--=-dj+xINrjE9lleBrubhQ/--