Intel about AMD

[mat]

http://www.theinquirer.net/modelperf.pdf

marketing... you just have to love it

 

[ronaldo]

Heh, Intel is starting to look a little desperate. :smile:

Some choice quotes:

"Marketing campaigns should educate customers about products.
However, Athlon* model numbers actually confuse your customers."

"As software evolves and benchmarks are updated, performance ratings become outdated."

"• MHz is a factual and consistent attribute of processor performance."

"• When MHz is combined with benchmarks that reflect your customers’ usages, they offer
a true assessment of performance."





<font size=-1>[ This Message was edited by: ronaldo on 2002-03-04 00:24 ]</font>

 

[bystander]

Looks as bad as the nVidia on Kyro document, or the ATi on GeForce 4 presentation.

Intel are obviously scared.

 

[demalion]

Well, though they ignore certain mathematical principles to artificially inflate the disparity between PR and MHz as the Athlon XPs are clocked higher, they are right that it isn't a linear progression and that the disparity grows.

OTOH, how about those benchmark results? I was aware that the Northwood @ 2GHZ did beat the XP 2000 at some benchmarks, but I think the quoted benchmark results are an effort at misrepresentation. Do they even use SSE on the XP, for example? The indications in the fine print are certainly that the Northwood was given every optimization they could think of.

Printing out something like that will probably be pretty effective on consumers, and will probably help salesmen sell some more SDR P4 systems. *sigh*

 

[Saem]

First of I haven't read the entire thing.

However, there are two quotes that were quoted here that are very important and big holes in the marketing numbers AMD uses.

"As software evolves and benchmarks are updated, performance ratings become outdated."

"• MHz is a factual and consistent attribute of processor performance."

They're completely right. When I put a processor into a machine, it runs at a certain MHz, it doesn't magically change. Marketing numbers become dated and can do so rather quickly due to patches, new revisions and so on. Something they don't consider. So, I'm pretty sure I don't aggre with all that's being said, I do believe that in those two cases, Intel is right on the money.

 

[davefb]

Wierd, i'd have said that those two "facts" meant Intel was exactly wrong!

Would a 2ghz 8086 be quicker than a 2ghz p4 ?


-dave-

 

[pcchen]

Well, if you think clock frequency is misleading, then the AMD's PR number is also misleading.

What the clock frequency tells us? Actually, only one thing is for sure: under the same circumstances, a processor with higher frequency will be as fast or faster than a processor, of the same kind, with lower frequency.

However, what is the AMD's PR number trying to tell us? If they just want to tell us "CPU with higher PR will be faster than CPU of the same kind with lower PR", they don't need to bring the PR number. Clock frequency will do the same thing. The purpose of the PR number is to compare between different kind of CPUs. Unfortunately, it is even more misleading than clock frequency. Comparing Athlon XP to classical Athlon may be ok for most cases. However, comparing Hammer to Athlon XP can be problematic, since their microarchitecture has some differences. Comparing Athlon XP to Pentium 4 is even worse.

The bottom line is, it is impossible to compare two different CPUs with just one number. And that's what the PR number is trying to do.


<font size=-1>[ This Message was edited by: pcchen on 2002-03-05 15:01 ]</font>

 

[Gubbi]

It is also the same Intel is trying to do with MHz. Just go compare a 1.8 GHz Northwood to a 2.0 GHz Willamette.

It's all marketing, real performance is measured by wall-time of your favourite apps.

Cheers
Gubbi

 

[mat]

http://www.theinquirer.net/06030203.htm

LOL!!