AMD Doomed?? PR Raiting Misleading??

[Nate420]

From AMD.com:

Q: What is this new P-rating ?

A: The P-rating is a new performance measurement, it is an apples to apples performance rating for the processor using the industry standard Winstone 96 benchmark. If an AMD-K5 processor has a rating of "PR100" that means that the processor would offer you performance equal to or greater than a Pentium at that P-rating. For Example, if you have a AMD-K5-PR133, it would give you the performance level of a Pentium 133MHz processor. However, the "PR" rating is not an indication of clock frequency.

 

[Emo]

<< From AMD.com:

Q: What is this new P-rating ?

A: The P-rating is a new performance measurement, it is an apples to apples performance rating for the processor using the industry standard Winstone 96 benchmark. If an AMD-K5 processor has a rating of "PR100" that means that the processor would offer you performance equal to or greater than a Pentium at that P-rating. For Example, if you have a AMD-K5-PR133, it would give you the performance level of a Pentium 133MHz processor. However, the "PR" rating is not an indication of clock frequency. >>



Nate, what you are quoting is AMD's old Pentium Rating (PR) system which AMD used with the K6 and other processors. Here is AMD's white papers on the Athlon XP rating system. If you read carefully, they provided us with a number of benchmarks between XP and Pentium 4 systems. However, they provide only one benchmark with the 1.4Ghz T-bird (Windows Media Encoder 7) on which they base the rating system. To be complete, those benchmarks should have included results with the 1.4Ghz T-bird which makes me wonder if some of those benchmarks were not favorable for AMD to show. BTW, my current desktop is an Athlon XP system, for all you AMD zealots.

 

[mee987]

I think the PR system may seem like a marketing trick by AMD at first, but if you look at all of the facts I think most of us would agree that it is very necessary. lets face it, if the athlon xp 1600+ (@1400mhz) was labeled as athlon xp 1.4, most people would just buy the t-bird 1.4. both are 1.4ghz, both will have equal performance, and the tbird is cheaper. tbirds rule! xp's suck!

I wasnt trying to start a flame war, I just wanted to point out how some of his seemingly logical predictions were not only strange but irrelevant. Im suprised he didnt try to say that since intel went from secc cpus to a smaller s370 to an even smaller 478 package that in 5 years the cpu's will be about the size of a sesame seed.

 

[Steppy76]

<< Oh yeah one more thing; it seems that other forum has a real Intel bias or something: the last post was talking about if Intel made up a "PR rating" of their own, like calling the P4 2.5 GHz "Performance 5000" or something, and the guy said we would cry bloody murder if they did. Damn right we would! Giving it a stupid name like that is totally misleading! My point is this: with Intel being basically the market leader, AMD designed its PR rating realative to Intel's speed. It is a direct performance comparaison to the Pentium 4; the name "Performance 5000" is not relative to anything.

These guys want to make themselves sound so informed (the guy talking about AMD getting good/bad yields on 2.2 GHz Hammers was one of the funniest!); they are just pulling this stuff out of thier @$$es. Let AMD worry about coming out with the Hammer CPU's; I'm sure they will thoroughly test them and come up with a PR rating that is more than fair in comparaison to Intel's flagship at that time. And if AMD skimps out and overshoots it with thier PR rating, we will always have Anandtech, Tom's Hardware, Sharky Extreme, and many other hardware sites to give us the skinny on just how well all of these next-gen platforms will perform. >>



That was my post, and if you would read it, I said nothing about calling it "the performance 5000". I just said if they gave them model numbers(ie the radeon 8500 isn't relative to anything either). Like calling a P4 2.5 a P4 model 5000. Nowhere did I indicate that it was a Performance rating(and if it WERE a PR rating, they could choose a 286 as the performance comparison, to make it as legit as the PR rating) And what benchmarks constitute the PR rating? Intel could take some benchmarks in certain media encoding stuff where it blows away an XP and give it a high "PR" based on that. Show me how it is any different than what AMD is doing.

If the PR system is supposed to indicate T-Bird performance, then the model system SHOULDN'T change when the P4 changes(like the FSB announced today). If they don't change it, we could very easily be in a situation where an Athlon ?? 3000+ will NOT perform as well as a P4 3000. Since this 3000+ IS compared directly to the 3000Mhz of the P4 in the real world regardless of what its "meant" to compare with, this doesn't seem the slightest bit shady to you? If they DO change the ratings to reflect P4 changes, then that basically means the the supposed "it's a comparison to T-Bird" argument has no merit. The PR ratings I think were fine for a while accuracywise, but have now put AMD between a rock and a hard place(the situation I outlined above).


If intel had ratings I'd think they were just as bad(I hate ratings from ANY manufacturer in the first place, not because of an Intel bias). I am sitting here typing this on my AMD XP 1600, so I don't think I'm too biased. I do however think Intel is sitting prettier with their tech than AMD is, which is MY OPINION...just because it differs from your doesn't mean it is biased.

 

[Sid03]

let me get this straight...

according to amd and the recently indicted (in the enron scandel) arthur andersen consulting, an athlonxp 2000+ is on par with the performance of a tbird 2000mhz (if one existed.)

since in many cases the xp 2000+ is equal to a p4 2000mhz, that would mean that a p4 performance is equal to a tbird clock for clock.

who here buys that a tbird 1600mhz is equal to a p4 1600mhz?

(PR = a bunch of crap)

 

[CoDerEd]

I wished amd stick with the clock for their CPU name, PR rating just make everybody confused.
people can see from the bencmark if the 1.6Ghz Athlon is faster or slower compare to 1.6Ghz P4.
there is a lot of site and review we can see about the performance, how well they OC and so.
Few of us that know nothing about computer and go to BestBuy, well the Sales should be well informed and
have to be fair to explained it to them about the performance difference as well they can see the price.

peace

 

[SSXeon5]

<<

<<

They arnt bias thats you

They are right .... MHz does mean MHz .... like i said before this is the greatest thing to happen to Intel ... a core that can clock to 10Ghz and stays at 6 IPC .... I want to see amd do that!

>>



I'm not biased against Intel; I own and love my P4! I'm just trying to get a point across. The PR rating system for AMD is necessary for non-techies like us.



<< Can people stop that "It seems too many people dont understands the pr rating a 1800xp is equal to a athlon (thunderbird model) running at 1.8ghz and NOT a 1.8 pentium" CRAP go get a 1.33Ghz XP (1500+) and a oced athlon 1.4@1.5 and the Tbird will rape it .... there is no way a 1.53Ghz XP can match a 1.8GHz Tbird

And where is the TBred .... o delayed again ..... and hammer .... 2003 .... ah ok >>



Your anti-AMD bias is evident though: a 1.8 GHz Tbird destroying a 1.53 GHz XP; let's compare actual CPU's, not made up ones please! >>



Read what sid said

.... and think before you speak Yet your defending AMD's stupid marketing scam ....

CoDerEd I agree with you 100% Just keep the Mhz title ... look at apples ... they dont use PR yet they are paster then P4's in alot of graphic benchmarks. They still know that Mhz means Mhz.


SSXeon

 

[SSXeon5]

<<

<< They arnt bias thats you

They are right .... MHz does mean MHz .... like i said before this is the greatest thing to happen to Intel ... a core that can clock to 10Ghz and stays at 6 IPC .... I want to see amd do that! >>



You've just hit on exactly why AMD needed the PR numbers in the first place. Everyone looks at P4s (even techies apparently) and drools over the high clock speeds. What many people don't seem to understand is that clock speed is a stupid way of measuring performance of a CPU. Faster does not always mean better, anyone who's ever worked with Sun or SGI boxes knows this. Unfortunitly, this is the real world and marketing plays a big roll, and most consumers look at one thing, how fast does the CPU run. Intel figured this out and made a CPU that plays to consumers. AMD was left with (at the time the P4 came out) far superior chips that no one would by because they were "slower". So they needed a bigger number to attach to their CPUs. >>



Hmm .... lets look at spec testing:



<< MISSION: To establish, maintain, and endorse a standardized set of relevant benchmarks and metrics for performance evaluation of modern computer systems. >>



SPEC CPU2000 Results:

CPU / Base / Peak

CINT2000

-P3 1.26Ghz_133Mhz i815 / 611 / 623
-P3 1.4AGhz_133Mhz i815 / 648 / 664
-P4 2.0AGhz_400Mhz i850 / 738 / 739
-P4 2.2AGhz_400Mhz i850 / 786 / 806
-P4 2.4AGHz_400Mhz i850 / 819 / 833

-XP 1.47Ghz_266MHz Epox 8KHA+ / 633 / 656
-XP 1.53GHz_266Mhz Epox 8KHA+ / 648 / 671
-XP 1.60GHz_266Mhz Epox 8KHA+ / 677 / 701
-XP 1.67Ghz_266Mhz Epox 8KHA+ / 697 / 724

CFP2000

-P3 1.26Ghz_133Mhz i815 / 422 / 440
-P3 1.4AGhz_133Mhz i815 / 437 / 456
-P4 2.0AGhz_400Mhz i850 / 744 / 764
-P4 2.2AGhz_400Mhz i850 / 779 / 801
-P4 2.4AGHz_400Mhz i850 / 806 / 812

-XP 1.47Ghz_266MHz Epox 8KHA+ / 561 / 604
-XP 1.53GHz_266Mhz Epox 8KHA+ / 572 / 615
-XP 1.60GHz_266Mhz Epox 8KHA+ / 588 / 634
-XP 1.67Ghz_266Mhz Epox 8KHA+ / 596 / 642




Look how well the P4 does ..... the athlon needs to get around 2.0GHz to match a 2.4A in CPU2000. And the XP needs to hit around 2.66GHz to match the 2.4A in CFP2000. Now for only 6 IPC the pentium 4 does amazing ...... In a few look for the 2.53GHz B's in there And look at the 1.4GHz Tualatin in CPU2000 it does better then the 1700+ (1.47GHz) and its lower clocked and half fsb. Thats interesting

SSXeon

 

[mee987]

those are synthetic benchmarks, check out a few of the anandtech articles that show real-world performance.

 

[mee987]

<< Few of us that know nothing about computer and go to BestBuy, well the Sales should be well informed and >>

you are kidding right?

 

[Nate420]

Thanks Emo...for the link

 

[archmage]

The PR Rating clearly does NOT translate into equivalent P3 or Tbird performance, no matter what AMD marketing claims.

The PR rating goes up 100 for every 66 Mhz of XP clock speed, but AMD has provided no compelling support for claiming that the XP scales 50% better than a Tbird or Tualatin. So even if you accept the claim that a 1.33 GHz XP is equivalent to a 1.5 GHz Tbird (or P3), the outlandish claim that a 1.73 GHz XP is equivalent to 2.1 GHz Tbird or P3 is not very credible.

PR ratings are clearly targeted at the P4, not the Tbird or the Tualatin.

 

[Priit]

Why whine about marketing? It has always been stupid (and for stupid people) and probably always will be. No matter if it's AMD, Intel or whoever, you shouldn't trust advertisement. So AMD's model ratings (BTW, it's NOT PR rating, SSXeon5) are meaningless random numbers and Pentium 3 doesn't make my internet faster....

 

[imgod2u]

Ok, what's with this 6 IPC thing? Neither the P4 nor the Athlon does anywhere near 6 IPC. 6 instructions per clock? Are you insane? The pipeline only lets out a MAXIMUM of 1 instruction per clock ASSUMING that the pipeline was completely filled and the instructions only required a maximum of 2 FPU (1 load/store and one add/multiply) and 4 ALU operations AND there wasn't a branch mispredict. CPU's work on 1 input, 1 output (except for SIMD but that's still 1 instruction, just multiple data results from 1 instruction). The Athlon does better AVERAGE IPC because when an error happens (branch mispredict), it wastes LESS time cleaning up the pipeline (10 cycles) as opposed to the P4 (20 cycles). There is also the reason that the Athlon has more FPU power therefore if an instruction requires 4 FPU operations (which the Athlon can complete in one pass) the P4 would need 2 passes to complete it, halting the pipeline. I think this was a mistake on Intel's part, they should've kept the FPU power along with adding the SSE/SSE2 (certainly possible with the Northwood die). There's also the problem of the P4 needing to fill that 20 stage pipeline (it tries to take in 1 instruction every clock and tries to output 1 instruction every clock, meaning in a 2.4 GHz P4, it needs to fetch 2.4 billion instructions every second to fill the pipeline while an Athlon at 1.73 GHz only needs to fetch 1.73 billion every clock, which is why it doesn't need as much memory bandwidth as the P4). IDEALLY, the P4 would have just as high IPC as the Athlon (provided enough execution units of course), since BOTH try to finish an instruction every clock (some instructions take more than a clock so no single-core CPU actually will reach the full 1 IPC) but again it's not an ideal world and branch mispredicts do happen (along with idle stages in case there isn't enough memory bandwidth or the code won't provide 2.4 billion instructions to put into the pipelines every second). This is where faster memory bandwidth, more cache, and particularly Hyperthreading comes in. If one thread of code doesn't provide enough instructions per second to fill the pipeline, another thread may fill in the gap (you still have the limitations if you have more instructions comming in than the processor needs). And the faster memory and more cache will help fetch enough instructions to fill that pipeline. AS LONG as the pipeline is kept filled and the instruction don't require repeated stages (in other words, something has to be redone, such as in the instance of the instruction requiring more than 2 FPU units in which it will take multiple passes to do) the P4 will have 1 IPC JUST LIKE the Athon would if its 10-stage pipeline was filled.

 

[shoez]

imgod2u has just laid it down straight. If anyone has actually studied how CPU works, even at a basic level, they will understand that the clock rating these days, is not a great performance indicator. When will people realise ?


tom

 

[dajeepster]

bah.. I own stock in both

 

[AznRyda]

<< imgod2u has just laid it down straight. If anyone has actually studied how CPU works, even at a basic level, they will understand that the clock rating these days, is not a great performance indicator. When will people realise ?


tom >>



never, because that what imgod2u posted is way to complicated for the average consumer.

 

[SteelCityFan]

In my opinion, AMD should have launched an ad campaign educating consumers on the Performance does not equal Mhz lesson... not make up a PR rating... the majority of their sales are to people who know what they are talking about since not too many OEM's sell their processors. I think this, coupled with the education of PC users would have done better than trying to make themselves appear to compete with Intel in Mhz.

And if some of you are gullible enough to believe AMD had no intentions of their PR rating linking to speeds of the P4, I may just be able to sell you the Statue Of Liberty.

They say it is linked to the TBird, but come on. Take the blinders off. As someone here pointed out, their pricing groups are pretty close when comparing Model Number Vs Intel's Mhz rating. You can't say it is equal to the performance of a 1.8Ghz TBird because none exist... sure they could estimate, but there are too many factors to make it unreliable.

 

[rogue1979]

AMD does not have the best PR department. But when it comes to computing power for the money spent, AMD has been the undisputed king for awhile now. Most computer geeks now this, the general public falls for the P4 higher MHz hype. There is nothing wrong with a P4 cpu, but simply put AMD is cheaper. Falling back to the old performance basis of the P3 and the Athlon, I still feel that the P4 is somewhat misleading in performance by the fact that they execute instructions slower than these old cpu's clock for clock. I mean c'mon, 8k of L1 cache? For the uneducated masses Intel hit the nail on the head, P4 sound faster than P3 and 1.5GHz sounds faster than 1GHz. Too bad Intels own $63 Celeron 1.0a overclocked will beat any Willy or Northwood running 1.8GHz or less. I guess Intel was really smarter to choose the route that made their cpu sound faster. And yes, a Northwood 1.6 or 1.8 will give an excellent overclock, but the general public will never know or use this. Bottom line, Northwood 2.2GHz $370 shipped on pricewatch, XP 2100 $220 on pricewatch. And for us geeks: While the 2.2GHz Northwood can be overclocked to be the king, in just a few short weeks the thorougbred will be released. My prediction is that the 1800 version will be less than $100 and go over 2GHz with air-cooling. So all you AMD non-believers get your digs in now, because the second it ships I will be here saying "I told you so!"

 

[shoez]

The best marketing department (note, not PR ) is owned by Intel, basically because they have huge cash reserves to spend on ad campaigns. the P4m and P4 are constantly advertised here in the UK, and especially during primetime, not to mention Intel having each company who uses their name to display the little Intel Inside ident. Intel, on the whole, win the consumer battle because they spend big, big bucks on advertising. AMD need to keep up somehow, and by *naming* their products along the line of Intel's, they can spread a little FUD


tom

 

[imgod2u]

<< AMD does not have the best PR department. But when it comes to computing power for the money spent, AMD has been the undisputed king for awhile now. Most computer geeks now this, the general public falls for the P4 higher MHz hype. There is nothing wrong with a P4 cpu, but simply put AMD is cheaper. Falling back to the old performance basis of the P3 and the Athlon, I still feel that the P4 is somewhat misleading in performance by the fact that they execute instructions slower than these old cpu's clock for clock. >>



It's only misleading if you don't understand processors. You can't blame Ford just because people buy their cars because they have a higher HP, even though their 0-60 may not be better than a Celica with a lower HP. Consumer stupidity is no reason to go against the product maker. Although I have to say Intel has milked this lack of knowledge among the consumers a lot more than I find tasteful.



<< I mean c'mon, 8k of L1 cache? >>



The P4 does NOT have "8k" of L1 cache. It has a trace cache capable of storing 12k of already decoded micro-ops. How this compares to normal caching varies considerably and people have only take guesses at what it would be equivalent to.



<< For the uneducated masses Intel hit the nail on the head, P4 sound faster than P3 and 1.5GHz sounds faster than 1GHz. Too bad Intels own $63 Celeron 1.0a overclocked will beat any Willy or Northwood running 1.8GHz or less. I guess Intel was really smarter to choose the route that made their cpu sound faster. And yes, a Northwood 1.6 or 1.8 will give an excellent overclock, but the general public will never know or use this. Bottom line, Northwood 2.2GHz $370 shipped on pricewatch, XP 2100 $220 on pricewatch. And for us geeks: While the 2.2GHz Northwood can be overclocked to be the king, in just a few short weeks the thorougbred will be released. My prediction is that the 1800 version will be less than $100 and go over 2GHz with air-cooling. So all you AMD non-believers get your digs in now, because the second it ships I will be here saying "I told you so!" >>



Looking at that overclocking test on the overclockers.com.au site, the T-bred seems to be maxed at 2.0 GHz and that was with a Vapchill system. Of course, they could've screwed the test up but I'd definitely say you shouldn't jump to any conclusions at this time.

 

[Busch]

maybe SSxeon needs to pay attention in school! what the hell is "raiting"? you misspelled rating twice

 

[rogue1979]

We understand the P4 capability here. A 2.5GHz-3.0GHz P4 kicks butt. To get 3GHz the 2.2 Northwood is still expensive, and the 1.6 and 1.8 Northwoods cost more than the XP 1700-1800 counterpart, which when both the Intel and AMD's are overclocked performance is about equal. Just pointing out that that Intel is more expensive, not putting down P4. I guess I am just old fashioned about progress and technology though, I still think that a newer better processor should be more efficient and faster clock for clock, not slower.

Well, seeing how this is AMD's first take at .13 micron cores, I don't think it is jumping to conclusions to assume that the yields will get much better real quick. Many people jumped to conclusions that because of die size and power output of the t-bred, it would have to run hotter than an XP. Well, with three t-breds to test back to back with the XP, those Aussies show the .13 micron core does run cooler. This is what we will have to believe until we actually get some in our hot sweaty hands, which should be real soon. I see no reason to believe the ones we get will run "hotter".

 

[ZimZum]

AMD and Intel are old news. Cyrix's new SkuzzleButt chip is going to smoke them all.

 

[imgod2u]

<< We understand the P4 capability here. A 2.5GHz-3.0GHz P4 kicks butt. To get 3GHz the 2.2 Northwood is still expensive, and the 1.6 and 1.8 Northwoods cost more than the XP 1700-1800 counterpart, which when both the Intel and AMD's are overclocked performance is about equal. Just pointing out that that Intel is more expensive, not putting down P4. I guess I am just old fashioned about progress and technology though, I still think that a newer better processor should be more efficient and faster clock for clock, not slower. >>



If you understood how processors work, you'll see that's not the whole idea at all. The point is to make a faster processor, whether by higher IPC or higher yield.



<< Well, seeing how this is AMD's first take at .13 micron cores, I don't think it is jumping to conclusions to assume that the yields will get much better real quick. Many people jumped to conclusions that because of die size and power output of the t-bred, it would have to run hotter than an XP. Well, with three t-breds to test back to back with the XP, those Aussies show the .13 micron core does run cooler. This is what we will have to believe until we actually get some in our hot sweaty hands, which should be real soon. I see no reason to believe the ones we get will run "hotter". >>



Well, they do run cooler but the overclocking results weren't exactly impressive. Cooling is not the only factor in processor yield.