Opteron vs. Nocona benchmarks

"Judd" wrote :

Why did they cripple the Intel box with 400 MHz DDR2? DDR2 right
now has a higher latency than DDR and won't perform as well at an
equal MHz. Why not use the 533 since that's what many of the
systems will ship with?

because its not supported ?


Pozdrawiam.
--
RusH //
http://pulse.pdi.net/~rush/qv30/
Like ninjas, true hackers are shrouded in secrecy and mystery.
You may never know -- UNTIL IT'S TOO LATE.

On Sat, 10 Jul 2004 19:21:07 -0600, "Judd"
wrote:
Why did they cripple the Intel box with 400 MHz DDR2? DDR2 right now has a
higher latency than DDR and won't perform as well at an equal MHz. Why not
use the 533 since that's what many of the systems will ship with?

Uhh, Intel's E7535 chipset doesn't support DDR2 533 memory, so that's
a pretty damn good reason why they wouldn't use it! These are
workstation/server chipsets, not the desktop i915/i925 chipsets where
talking about here, different requirements and different specs. Both
systems were equipped with the fastest available setups.

Besides, given the shared-bus nature of the Xeon chips, the difference
between DDR2 400 and DDR2 533 is likely to be rather minimal, probably
in the 1-2% range.

-------------
Tony Hill
hilla underscore 20 at yahoo dot ca

On Sun, 11 Jul 2004 10:12:41 +1200, Paul Gunson
wrote:

George Macdonald wrote:
Yeah but they didn't use it AFAICT. IOW they completely missed the point
of the whole exercise by running 32-bit software and not actually showing
their hand very clearly. D'oh - it's the 64-bit comaprison we want to see;
the 32-bit stuff has been run on previous Opteron/Xeon comparos months ago
- nobody cares now.

but they havn't run comparisons with 800Mhz FSB Xeons before have
they...? even thought it was 32 bit apps on XP32 (i think?) i still
found the benchmarks very useful - as a maya user looking to buy the
best performing hardware for my render farm now, when 64 bit maya
arrives a year [or 3] from now. 32 bit banchmarks are better than no
benchmarks.

I haven't examined the Xeons in detail but has nobody ever overclocked one
to a 400MHz FSB? Even so, P4s have been there for a while so I don't see
anything particularly new here. If you're running dual Xeons at 266MHz FSB
I'd think you have a good idea how much you are losing on memory
performance vs. 400MHz. If "now" is annual splurge time, I see your point
though; OTOH if you keep systems for even 2 years, I think you should pay
close attention to the 64-bit comparisons... when they become available.

The fact is that the article *is* misleading with its "64-bit Battle" label
where there is not a single mention of a 64-bit benchmark amd 32-bit OS
*was* used. I don't think I'm alone in being intrigued by Intel's err,
coyness on 64-bit performance... and the resounding silence on the Web on
the subject. The few dribbles we've seen suggest a possible disaster for
Intel... and the mention in the article of "quick 64-bit SiSoft Sandra
benchmarks" with no actual numbers could be ominous.

BTW if Maya is looking at =1year for 64-bitness I'd say it's time to ring
their bell.:-)

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

On Sat, 10 Jul 2004 19:21:07 -0600, "Judd" wrote:


"George Macdonald" wrote in message
news
On Sat, 10 Jul 2004 04:36:10 GMT, "Yousuf Khan" wrote:

RusH wrote:
"Yousuf Khan" wrote :

seems to be better at
workstation-style applications than at server-style applications.

like every Xeon vs Opteron comparison. Why ? Windows XP Professional
Edition.
ROFL

Why should that make a difference?

Oh and BTW, it looks like these GamePC people have been able to get XP64
working on the Nocona. Either they have the latest public beta, or they
have
a private beta.

Yeah but they didn't use it AFAICT. IOW they completely missed the point
of the whole exercise by running 32-bit software and not actually showing
their hand very clearly. D'oh - it's the 64-bit comaprison we want to
see;
the 32-bit stuff has been run on previous Opteron/Xeon comparos months ago
- nobody cares now.


Why did they cripple the Intel box with 400 MHz DDR2? DDR2 right now has a
higher latency than DDR and won't perform as well at an equal MHz. Why not
use the 533 since that's what many of the systems will ship with?

The Xeons and 7525 support 400MHz FSB & memory - I dunno if it's planned to
go to 533MHz.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

On Sun, 11 Jul 2004 02:26:38 -0400, George Macdonald
wrote:
The Xeons and 7525 support 400MHz FSB & memory - I dunno if it's planned to
go to 533MHz.

Nocona with any of the three Lindenhurst MCHs uses an
800mega-transfers-per-second fsb, and dual-channel
400mega-transfers-per-second memory interconnects.

Lindenhurst/Tumwater MCHs support slower memory operation - even "DDR1"
memory, if configured for such. Intel has roadmaps extending through DDR2-800
(though not for this particular chipset family), you can probably Google one
up...

/daytripper (Agent wants to change "Tumwater" to "Dumbwaiter". Hmmm....)

"Ed" wrote in message
...
On Sat, 10 Jul 2004 19:21:07 -0600, "Judd"
wrote:


Why did they cripple the Intel box with 400 MHz DDR2? DDR2 right now has
a
higher latency than DDR and won't perform as well at an equal MHz. Why
not
use the 533 since that's what many of the systems will ship with?


what difference would it make? 3% tops?

Is 3% a small difference? I'd certainly take it. I understand that it's
not supported. I didn't realize that.