Date: Sat, 8 Jul 2000 22:54:53 +0200 From: Stefan Esser <se@freebsd.org> To: Paul Murphy <pnmurphy@home.com> Cc: Joe Greco <jgreco@ns.sol.net>, "Jordan K. Hubbard" <jkh@zippy.osd.bsdi.com>, ler@lerctr.org, grog@lemis.com, Greg@fatcanary.com.au, stable@FreeBSD.ORG, Stefan Esser <se@freebsd.org> Subject: Re: AMD K6-2 / 550 Message-ID: <20000708225453.E2104@StefanEsser.FreeBSD.org> In-Reply-To: <3962A197.62560B50@home.com>; from pnmurphy@home.com on Tue, Jul 04, 2000 at 10:46:47PM -0400 References: <200007050229.VAA40330@aurora.sol.net> <3962A197.62560B50@home.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2000-07-04 22:46 -0400, Paul Murphy <pnmurphy@home.com> wrote:
> Is there a particular reason for going with the AMD's over Intel? When
> buying computers I have always stayed with the Intel CPU's because I
> thought it was similar to Soundblaster vs. others: all the others say
> they are "Soundblaster compatible" so why not buy the real thing?
Perhaps because the Soundblaster may have been the first to use some
(low level) programming interface, but may not be the best hardware.
I always avoided Creative products, since they just didn't seem as
good as other companies offerings (except for products developed by
companies they bought and rebranded, like Ensoniq PCI sound-cards).
But with regard to using the AMD K6 in preference to Intel chips:
There was a time when Intel considered details about their processors
a trade secret, and it was not possibly to write a free compiler for
the Pentium, that knew how to make good use of the second pipeline.
There were very complex interdependencies. It often sped up a loop
considerably, if a few NOPs were inserted at the right places, which
is counterintutive, to say the least ;-)
That is the reason, that the AMD K5 performed so much better under
FreeBSD than the Pentium it was rated against. The K5 had an execution
engine much like that later introduced by Intel in the PentiumPro (i.e.
what every current i586 and up compatible chip except for VIA/Cyrix/IDT
does: pre-process the i86 instructions into RISC instructions for multiple
independent execution units). For that reason, the K5 did not depend on
such specific optimizations as the Pentium to keep its pipeline going.
With the (in)famous Appendix H being withheld from the GCC programmers,
the Pentium only reached some 70% to 80% of its nominal performance ...
If reliability (i.e. 7x24, year after year) is your primary goal, then
choose whatever seems to be well supported and proven. But for a home
system or non-critical workstation, I'd use what provides the best value
for the money. And that has often been an AMD processor in a motherboard
that had been declared obsolete by Intel years ago ;-)
(In fact, I know a number of people running a K6-3 in their five year
old Triton based mainboards. They easily beat my non-overclocked 300MHz
Celeron A).
Regards, STefan
PS: Anybody seen any indications, that the K6-2+ or K6-3+ may become
available as an upgrade processor ? Its low power consumption
should make it ideal for old mainboards with an linear regulators.
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000708225453.E2104>