Now AMD follows Intel's lead

[Black Jack]

AMD may begin adopting Intel's model numbering scheme for its own
Athlon 64's. Strange, AMD adopts model numbers and then Intel follows,
and then AMD retweaks to fit Intel. Well, so much for the PR-rating
system being an indicator of relative performance from an Athlon
Thunderbird.

http://www.xbitlabs.com/news/cpu/display/20040331173218.html

 

[Grumble]

AMD may begin adopting Intel's model numbering scheme for its own
Athlon 64's. Strange, AMD adopts model numbers and then Intel follows,
and then AMD retweaks to fit Intel. Well, so much for the PR-rating
system being an indicator of relative performance from an Athlon
Thunderbird.

http://www.hardware.fr/news/imprimer/6484/

The slide from AMD's presentation says:

Since AMD Processors P-Rating has always be a meant [sic] for
us to position our products in comparison with Intel processors
and since Intel has decided to give up their CPU branding based
on the frequency, we have also to move towards a new P-Rating
system. It is mandatory in order to let our customers understand
how we position our processors. These P-rating will be used from
third quarter 2004 and will concern only the Athlon 64's processors.

Prescott LGA775 @ 3.6 GHz = P4E 560
Prescott LGA775 @ 3.4 GHz = P4E 550
Prescott LGA775 @ 3.2 GHz = P4E 540
Prescott LGA775 @ 3.0 GHz = P4E 530
Prescott LGA775 @ 2.8 GHz = P4E 520

Athlon 64 3400+ (2.2 GHz ??) = Athlon 64 560+
Athlon 64 3200+ (2.0 GHz ??) = Athlon 64 550+
Athlon 64 3000+ (1.8 GHz ??) = Athlon 64 540+
Athlon 64 2800+ (1.6 GHz ??) = Athlon 64 530+

2800+ could be 1.6 GHz + 1 MB L2 or 1.8 GHz + 512 KB L2 (Newcastle)
3000+ could be 1.8 GHz + 1 MB L2 or 2.0 GHz + 512 KB L2 (Newcastle)
3200+ could be 2.0 GHz + 1 MB L2 or 2.2 GHz + 512 KB L2 (Newcastle)

Thus, in AMD's mind:

Prescott @ 3.6 GHz = Athlon 64 @ 2.2 GHz
Prescott @ 3.2 GHz = Athlon 64 @ 1.8 GHz (perhaps 2.0 GHz)

 

[Grumble]

AMD may begin adopting Intel's model numbering scheme for its own
Athlon 64's. Strange, AMD adopts model numbers and then Intel follows,
and then AMD retweaks to fit Intel. Well, so much for the PR-rating
system being an indicator of relative performance from an Athlon
Thunderbird.

http://theregister.co.uk/content/51/36722.html

:-o

 

[Never anonymous Bud]

http://theregister.co.uk/content/51/36722.html0

Got this far without twigging? Yes, it's an April Fool's gag, as the 1 April
2004 date on the original Chinese story reveals, but the French, reading the
story late at night on 31 March didn't spot. Nor did a number of US sites who
spotted the Hardware.fr piece early in their timezones, but long after it had
already become April in the Far East.






To reply by email, remove the XYZ.

Lumber Cartel (tinlc) #2063. Spam this account at your own risk.

This sig censored by the Office of Home and Land Insecurity....

 

[Rick Jones]

Drift...

Given the number of times the old guard in comp.arch is able to
describe how some "new thing in computing" being touted somewhere or
other was done N years ago on system bar or OS foo, I have to wonder
if despite various peoples' desire to anoint "leaders" and "followers"
the proper description of the industry is really closer to "Duck Duck
Goose" or "Ring Around the Rosy" than "Follow the Leader."

rick jones

 

[Nadeem]

Drift...

Given the number of times the old guard in comp.arch is able to
describe how some "new thing in computing" being touted somewhere or
other was done N years ago on system bar or OS foo, I have to wonder
if despite various peoples' desire to anoint "leaders" and "followers"
the proper description of the industry is really closer to "Duck Duck
Goose" or "Ring Around the Rosy" than "Follow the Leader."

rick jones

damn, that's what i call a long sentence!!! ;-)

--

Nadeem M Nayeck [ m n n a y e c k @ i n t n e t . m u ]
______ ______ .
.:_\_ . \\_ . \_::.
. .::./ ./ // ./__/.:::. . Registered LU #290695
:_<_____/<____ >_:.
. \/ .

 

[Carlo Razzeto]

AMD may begin adopting Intel's model numbering scheme for its own
Athlon 64's. Strange, AMD adopts model numbers and then Intel follows,
and then AMD retweaks to fit Intel. Well, so much for the PR-rating
system being an indicator of relative performance from an Athlon
Thunderbird.

http://www.xbitlabs.com/news/cpu/display/20040331173218.html

Actually in a strange way this exactly what AMD wanted from the start.
Universal model numbers for all processors that describe their relitive
preformance to eachother better than MHz. I guess in a weird way it's an AMD
win still.

Carlo

 

[The little lost angel]

Universal model numbers for all processors that describe their relitive
preformance to eachother better than MHz. I guess in a weird way it's an AMD
win still.

The question is, will the consumers win? I can envisage a situation
that goes like this

1 May 200x : Intel announces x.1 Ghz Model 810 CPU
3 May 200x : AMD announces a.1 Ghz Model 810+ CPU
5 May 200x : Intel annnounces x.15 Ghz Model 812 CPU
7 May 200x : AMD announces a.15 Model 815+ CPU
9 May 200x : Intel announces x.2 Ghz Model 820 CPU
11 May 200x :.... and so on and so forth...

pP


--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

 

[Carlo Razzeto]

The question is, will the consumers win? I can envisage a situation
that goes like this

1 May 200x : Intel announces x.1 Ghz Model 810 CPU
3 May 200x : AMD announces a.1 Ghz Model 810+ CPU
5 May 200x : Intel annnounces x.15 Ghz Model 812 CPU
7 May 200x : AMD announces a.15 Model 815+ CPU
9 May 200x : Intel announces x.2 Ghz Model 820 CPU
11 May 200x :.... and so on and so forth...

pP


--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

Is the customer really winning with the old MHz is the only way to measure
preformance method? For Intel customers looking at laptops it means they
were buying P4-M systems when it's really not the best mobil solution out
there...

Carlo

 

[The little lost angel]

Is the customer really winning with the old MHz is the only way to measure
preformance method? For Intel customers looking at laptops it means they
were buying P4-M systems when it's really not the best mobil solution out
there...

Definitely, I don't agree that the Mhz was that fantastic a measure of
actual performance, but this entire model number thing is going to
murk up the waters even more because the marketing dept is just going
to have a field day with it and this time, they don't even have to ask
the engineering department a single question!

If they could get away with it, they would probably sell a Model 580
and Model 585 that is no difference except for a CPU ID that tells the
BIOS I'm 580 and the other guy is 585

At the very least with an actual measureable metric like Mhz or a
equivalence rating, we make a good guess within the same architecture.
After a few models, it's also quite easy to establish X Mhz of A is
around Y Mhz of B in a general ratio of c.xx : d.xx

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

 

[Carlo Razzeto]

Definitely, I don't agree that the Mhz was that fantastic a measure of
actual performance, but this entire model number thing is going to
murk up the waters even more because the marketing dept is just going
to have a field day with it and this time, they don't even have to ask
the engineering department a single question!

If they could get away with it, they would probably sell a Model 580
and Model 585 that is no difference except for a CPU ID that tells the
BIOS I'm 580 and the other guy is 585

At the very least with an actual measureable metric like Mhz or a
equivalence rating, we make a good guess within the same architecture.
After a few models, it's also quite easy to establish X Mhz of A is
around Y Mhz of B in a general ratio of c.xx : d.xx

<snip signature>

Marketing would probably have to ask the engineers all kinds of questions
assuming neither Intel nor AMD want to pull a Cyrix (ever use a P-150+ or
contemporary processor)? They would have to know how cache size, bus speed
MHz and core refinements interact to affect performance.

Carlo

 

[Tony Hill]

Definitely, I don't agree that the Mhz was that fantastic a measure of
actual performance, but this entire model number thing is going to
murk up the waters even more because the marketing dept is just going
to have a field day with it and this time, they don't even have to ask
the engineering department a single question!

If they could get away with it, they would probably sell a Model 580
and Model 585 that is no difference except for a CPU ID that tells the
BIOS I'm 580 and the other guy is 585

I don't think we'll have to worry TOO much about that. We don't
generally see this with other products marketed using model
name/numbers (and that is virtually every product in the world) for
the simple reason that it quickly becomes obvious that the marketing
department is trying to pull a quick one on consumers. Consumers
really don't like this and tend to react rather negatively, so such a
move just isn't worth it.

At the very least with an actual measureable metric like Mhz or a
equivalence rating, we make a good guess within the same architecture.

The problem is that only a small percentage of the buying public know
enough about the architectures to tell one from the other. How many
of the buying public know that the Celeron 1.4GHz was a TOTALLY
different (and MUCH faster) chip than the Celeron 1.7GHz? Probably
about 1%. It's a little bit better with the P4 line, though there are
still some tricky parts there, ie the P4 3.0GHz being faster than the
P4 3.2GHz. It's also rather complicated on the mobile front where
there is a "Mobile Pentium 4" and a "Pentium 4-M" that had VERY
different power consumption characteristics (though similar
performance).

In short, I like model numbers, because at least everyone KNOWS that
they're a meaningless indicator of performance. MHz was pushed as an
important measure of performance when really it wasn't.

 

[RusH]

questions assuming neither Intel nor AMD want to pull a Cyrix
(ever use a P-150+ or contemporary processor)?

Cyrix P120 was as fast as P200MMX in RC56 crunching (weird) so its not
an obvious example.



Pozdrawiam.

 

[The little lost angel]

Marketing would probably have to ask the engineers all kinds of questions
assuming neither Intel nor AMD want to pull a Cyrix (ever use a P-150+ or
contemporary processor)?

I was a very happy user of my Cyrix 150Mhz PR 233 or something like
that. It did everything I wanted it to do in general and was much
cheaper than what the equivalent Pentium was.

they would have to know how cache size, bus speed
MHz and core refinements interact to affect performance.

They do? Ok maybe they do, but do they really care? I certainly don't
remember seeing any marketing materials describing in any details that
bigger memory bandwidth doesn't help with latency bound applications.
pppP

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

 

[Carlo Razzeto]

Cyrix P120 was as fast as P200MMX in RC56 crunching (weird) so its not
an obvious example.



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.

You really can't just take once benchmark and declare the chip a winner, I'm
sure that there are certain situations where these chips were as good or
better than their Intel counter parts... However in terms of general usage
these chips simply could not live up to their name/model number. Not like
the AthlonXP/64's have been able to more recently. This is one of the big
reasons why Cyrix ended up going under, they never abandoned PR ratings and
they also never really improved the preformace of thier chips (relitive to
the pentium it was competing against).

Carlo

 

[Carlo Razzeto]

I was a very happy user of my Cyrix 150Mhz PR 233 or something like
that. It did everything I wanted it to do in general and was much
cheaper than what the equivalent Pentium was.

<snip>

I wasn't very happy at all with my Cyrix PR-150+. It was ok I suppose,
usually better than my old P75... But I wasn't really happy until I got my
K62.. I never used the later Cyrix chips, but I really never felt the need
to after switching to AMD. Those chips went clock for clock with the
pentiums on the integer unit, and they were acceptable for gaming at the
time.

Carlo

 

[Grumble]

The problem is that only a small percentage of the buying public
know enough about the architectures to tell one from the other.
How many of the buying public know that the Celeron 1.4GHz was a
TOTALLY different (and MUCH faster) chip than the Celeron 1.7GHz?
Probably about 1%. It's a little bit better with the P4 line,
though there are still some tricky parts there, ie the P4 3.0GHz
being faster than the P4 3.2GHz.

Is that Northwood (P4C) @ 3.0 GHz vs Prescott (P4E) @ 3.2 GHz?

 

[Black Jack]

Actually in a strange way this exactly what AMD wanted from the start.
Universal model numbers for all processors that describe their relitive
preformance to eachother better than MHz. I guess in a weird way it's an AMD
win still.

I'm not sure why anyone would want or need universal performance
numbers to describe the chips from one maker to another? BMW and
Mercedes don't line up their model numbers just to make it easier for
consumers to compare their models between each other. BMW model
numbers are comparable only to other BMW models, and ditto for Merc
vs. Merc. At this point it is upto the salesman of the computer
systems to tell consumers that AMD X processor is competive to Intel Y
processor.

Yousuf Khan

 

[Nate Edel]

numbers to describe the chips from one maker to another? BMW and
Mercedes don't line up their model numbers just to make it easier for
consumers to compare their models between each other. BMW model
numbers are comparable only to other BMW models, and ditto for Merc
vs. Merc.

Actually, they're pretty close in several major classes:
Compact/low-end luxury: C-class to 3-series 4 door
Roadster: SLK to Z3
Mid-size luxury: E-class to 5-series
Full-size luxury: S-class to 7-series

At this point it is upto the salesman of the computer systems to tell
consumers that AMD X processor is competive to Intel Y processor.

Or general market positioning.

 

[Carlo Razzeto]

I'm not sure why anyone would want or need universal performance
numbers to describe the chips from one maker to another? BMW and
Mercedes don't line up their model numbers just to make it easier for
consumers to compare their models between each other. BMW model
numbers are comparable only to other BMW models, and ditto for Merc
vs. Merc. At this point it is upto the salesman of the computer
systems to tell consumers that AMD X processor is competive to Intel Y
processor.

Yousuf Khan

Actually, if you remember back to when AMD was first coming out with model
numbers for the AXP line they did bring up that they really wanted universal
performance ratings. I believe they even talked about setting up an industry
wide consortium for this purpose. I'd look up links but I'm too tried/lazy
to do that right now.

Carlo