AMD's Model Numbers Revisited

[doctj]

AMD's PR rating has been a success where it matters most, namely by enabling them to keep their average selling price (asp) up. Instead of pricing their 1.73 Ghz (2100+ XP) cheaper than a 1.7 Ghz P4 they only have to make it cheaper than a 2.2 Ghz P4, I think it was an excellent business move. I heard that the hammer will be introduced as model 3000+ which may be before Intel releases its own 3 GHz chip. Tech savvy people like us may not like it but we don't make up the majority of chipmaker's business.

[Rugor]

In many ways Intel is responsible for AMD's current rating scheme because they cheapened processor speed and raw Megahertz.

Before the Pentium 4, Intel's primary goal in introducing a new processor (other than to make money) was to increase the new processor's performance. Increasing speed was important, but only as part of the overall goal of producing a better performing processor. This can clearly be seen in the Pentium III line where the PIII coppermine not only scaled higher in speed than the previous katmai version, but also outperformed it at the same speed.

However the P4, despite significantly higher clockspeeds does not outperform the PIII in the same fashion. Every aspect of the P4's design is subordinated towards the goal of achieving higher clockspeeds. The simple fact that they have demo'd Northwoods at 3GHz+ using air cooling as well as recent overclocking results showed that they have succeeded brilliantly at that goal. In sheer clockspeed the P4 is screamingly fast. Even the weakened FPU and reliance on SSE2 has its roots in the quest for speed: it's a lot faster to make an SSE2 function call than to perform a complex calculation. Rambus too is all about going fast, it clocks much higher than SDRAM of any variety and trades off latency for speed.

A result of this is that Intel's marketing strategy has also shifted from "Upgrade to a system with our new processor because it has higher performance," to "Upgrade to a system with our new processor because it runs faster." Megahertz sells so more Megahertz should sell more.

AMD is still working on the previous paradigm: They want you to use the new processor because it outperforms its predecessor. But because of Intel's dominant position in the marketplace they had to make it look like they were shifting the same way Intel was.

It's like comparing apples and oranges when the buyer only thinks in terms of appleness. So AMD's had to come up with telling the public how many apple equivalents their orange is worth. We can go on all day about whether those equivalents are against blue or green apples (Intel or AMD processors), but the fact is that the performance rating system is supposed to be an approximate equivalent to competitive Megahertz.

AMD had to go to the Performance rating scheme because of how Intel designed and marketed the P4. Sadly they will always be second banana as long as Intel controls public perception.

[wallie_x]

It is not only pubic perception, but also political backing. Here in America Intel can do no wrong. ZDNET and CNET two big web purveyors of emergent technology are in bed with big money, namely Intel and Micro$oft. Big money capitalism has a tendency of supporting one another since it is in their best interests to do so. Intel rules the advertising media's airwaves. As I said in another post, the media in America has arrogantly taken on the task of attempting to socialize the populace into a perspective. The news media in particular has mutated into a 'Big Brother' attitude of” We know what's best for you poor unenlightened peons" so we will dictate social mores'. The whole capitalist machine works off a view that the general public is ignorant; a commodity to be exploited, even (if necessary) by a campaign of disinformation. Intel's behavior only substantiates this supposition.

[$1500-P4 gamer]

Ask yourself this though.What is a PR rateing anyhow?Notheing but a guess.My P4 1.5 on sisoft gets a PR rateing of like 1800mhz.Now I know that is wrong but there it is.So that rateing system is junk anyhow.The whole mhz rateing is junk aswell.The new chip archetectures make it impossible to define a chips speed through mhz.In the old days this was the easiest way to tell if you upgraded or not.This is not the case anymore.They should just make names for each chip and leave it at that.

The 486 changed as to add a math co-procesor to its core.

Then came the P-PRO which simply added the mmx tech over the pentium.

Next came the P2 which was simpley a PPRO on a smaller core and less core volts yielding less heat= higher mhz.

Then the P3 which added sse and smaller core again as to reach higher speeds.

All revisions between Pentium pro and p3 are the same core but smaller with subtle changes.Notheing deserveing of a entirely new nameing scheme.Intel has played it like it is.Would it have been better if the made it up to look as if it where more of a change than it really is?

Neither co' is dueing it right as mhz are out for measureing the speed of a cpu.And I still say AMD is only trying to fool the less informed pc enthusiest with there rateing system.Watch a infomercial where they are selling a AMD sys.They say that a XP 1600+ beats a P4 2.2A.That is a lie out and out,they are saying this to sell the unit and that is it.If they showed all the benches this would prove a person as to THINK about the #'s before buying.It is easier to confuse someone and make outlandish claims with a 1600+ rateing then useing the same rateing system as Intel.

Haveing said that I like both chips so don't be slaming for that last statement.I just don't think that either is that much better to be worth argueing about.

[Rugor]

Gamer, once again you made me think-- but here's my two cents back to you:

The Sisoft Sandra PRating is not the same as AMD's model numbering scheme. A Celeron 300A at 450MHz had a Sisoft rating of 541, and my XP 1800+ rates 2243 according to that program.

Since my Athlon and your P4 run at essentially the same speed let's look at a couple of things. Feel free to flame me if anyone thinks 33MHz is a big difference at 1.5GHz. According to Sandra the Athlon's performance is about 1.2 times that of the P4. Now comparing speed to model number, we find that the Athlon's model rating is 1.2 times both its own clock speed, or the clock speed of the P4. Based on this it can be argued that the AMD model number system does give at least a reasonable approximation of a comparative performance metric.

It all boils down to one problem: There is no real performance metric today. We need a way of keeping score so people can easily compare AMD and Intel chips (or even VIA and possibly SiS). Given that there's no real metric, AMD's model numbers represent the best solution they could come up with. It may not be a very good solution, but it's better than some.

PS: Pentium with MMX is just PentiumMMX, the original P-Pro doesn't have MMX, the PII introduced MMX to the P6 core.

[hacker15]

I agree up to a point - I just finished putting together an XP 1900+ based box on a Shuttle mobo with the new Via 233A chipset. I put in a gig of DDR PC2100 and a 40 Gb WD 7200 rpm hd, along with a GeForce 4 Ti. Guess what? It don't run much faster than my old Athlon TB 1.0 on an FIC SD11 mobo, with 768 of PC133 and two IBM Deskstar 20 Gb 5400 rpm hd's and the GeForce 2 I was using....

So much for "advancements in speed". It's kind of useless when you really can't see it - but it sure is nice to have the hardware.....

[wallie_x]

I agree with U P4 gamer in that there is no standard metric in which to judge a CPU's performance any more. The reality is that any company will use whatever tactic they can to position their selves to their own advantage. Both AMD and Intel use tactics that are designed to help them get a point across to the public, which both companies view as basically ignorant. The only thing I can say subjectively is that my Athlon XP, DDR, Windows XP machines is my dream come true. It has run all the software and peripherals I've hooked up to it very well. Gone are the days of driver incompatibilities and system freezes and endless searches on the web for fixes to problems I don't even understand. I have never extensively used a P4 machine, but I would like to for comparison sake. All I know is that my current machine's performance is excellent and for that I am happy.

Addendum: I would also like to see more specialization. We've seen a trend that some software attempts to be all things to all people, e.g. Windows Media Player. In attempting to take on such a broad spectrum of capabilities it succeeds in being mediocre in most of them. Why doesn't Intel push the P4's obvious strengths, its superior memory bandwidth which a gamer such as your self can appreciate much more than a photo nut like me who needs instructions per clock cycle to edit my photos? Instead of the posturing we see now days as one CPU being superior in all aspects to the other. Pax wallie_x

[davesino]

Originally posted by gvenditto
I always felt Intel screwed up the whole chip naming thing when they changed from the x86 system after the 486. Some marketing geniums came up with the Pentium name -- changing the CPU name from what it should have been called, the 586.

Intel knew they couldn't keep up the Latin-number thing after Pentium (for 5) because what would the name for the 6th chip be? The Six-ium ??

By my count, they would only be up to a 786 by now if they had just let nature take its course.

They actually got the Pentuim name from the Greek work for "five".

[davesino]

Originally posted by gvenditto
I always felt Intel screwed up the whole chip naming thing when they changed from the x86 system after the 486. Some marketing geniums came up with the Pentium name -- changing the CPU name from what it should have been called, the 586.

Intel knew they couldn't keep up the Latin-number thing after Pentium (for 5) because what would the name for the 6th chip be? The Six-ium ??

By my count, they would only be up to a 786 by now if they had just let nature take its course.

They actually got the Pentuim name from the Greek word for "five".

[davesino]

Originally posted by gvenditto
I always felt Intel screwed up the whole chip naming thing when they changed from the x86 system after the 486. Some marketing geniums came up with the Pentium name -- changing the CPU name from what it should have been called, the 586.

Intel knew they couldn't keep up the Latin-number thing after Pentium (for 5) because what would the name for the 6th chip be? The Six-ium ??

By my count, they would only be up to a 786 by now if they had just let nature take its course.

They actually got the Pentium name from the Greek word for "five".

[Strawbs]

Talk about rubbing ones' nose in it.
edit - delete
Talk about rubbing ones' nose in it.
edit - delete
Talk about rubbing ones' nose in it.

[Rugor]

Ummmmmmmm Pentium for 5

Hexium for 6

Heptium or Septium for 7

Octium for 8...........................

[$1500-P4 gamer]

Rugor-

"PS: Pentium with MMX is just PentiumMMX, the original P-Pro doesn't have MMX, the PII introduced MMX to the P6 core."

I have had both(and still do in the liveing room) and come to think of it you are right.On the PentiumMMX that is all it says.And the PPro chip I have has no mmx.That is why the 233mmx was better.So it went P-Pro then Pentium MMX then P2@66mhz fsb right?From there we all know it went to P2@100fsb-P3@100fsb then P3@133fsb. and now P4@100fsb x4.

[Rugor]

Gamer,

yup

[dosmastr]

Originally posted by zergboy5
rlbogue, I do belive that Tripleshot is correct when it comes to the numbering of the chips because the numbers were based off the micro archatecture and not the revision of that archatecture. so the number actuly go more like this:

Pentium = 586
Pentium Pro = 686a
Pentium II = 686b
Pentium III = 68c
Pentium 4 = 786

Only with the pentium 4 has intel relesed a truly revised desigen in sted of just tweaking the pentium pro archatecture.

He does have a point, really all pentiums except for original and the new one have been based on the same core
the only difference between the 2 and original 3 was the addition of KNI, katmai new instructions, sometimes dubbed MMX2. (intel noticed that people really didn't care about MMX, so they found a better way to make them think twice about how much better its sucessor would be the second time around-- they renamed the chip)

As for all those people who kept ranting and raving about how the model rating system wouldn't work, i lean back in my chair and scream I TOLD YOU SO!!

Cyrix and amd's orignal rating scheme was really..... generous, in many apps the 585 cyrix really was no faster then a 486. Amd learned and was much much more conservative with this new system. I for one would've liked to have seen what intel could have done if they had kept the p6 archetecture going, moving it to .13 micron put it withing striking distance of thier new p4 flag even (if you overclocked the thing anyway.... of course they wouldn't release a stock clock chip able to hurt the new king)
something you might note: pentium-- k5, pentium 2 k6-1/k6-2 p3 k6-3, p4-- k7 amd still seems to remember the old number scheme too, they are right on track they had 686b vs 686-2
and now they have 786 vs 786.
and for those of you who would still debate the model rating system with me, how about you look at statistics, never before has amd made as much profit per flagship sold then it has since not too long after the release of the XP core. Never before has the per chip price exceede the price per tray ( i am not certain what they mean by that but i did read it on a tech news site, and the site was talking about this same topic)

davesino, if you hit back on the browser, you have to hit it 3 times fast or you get that happening.... plus who didn't know the pent-- as in pentagon, didn't mean 5?

as for reality:
$1500-P4 gamer, dude you hit it right on the head.
the chip itself hasn't changed but with newer and more advanced chipsets by via and others athlon performance has increased 25% since introduction, even running on the exact same chip--- according to dr tom on tomshardware.com
also, look at benchmarks, some the p4 northwood 1.6 beats the XP 2100+, some the 2.2Northwood cant catch the athlon XP 1600
it depends what you plan on using the thing for, i'm sure if you averaged benchmarks it would be pretty close to the ratings but still who is "average" in the programs they use? video editors go for intel, SSE2 packs a whollop.

lol.... i really should start a website... where i can rant and rave like this...