AMD's PR system is now totally useless!

[Priit]

Both Mhz rating system and operating frequency are really totally useless indicating application(s) performance, IMO. Those numbers are only important for marketing (by obvious reasons). It's like when car salesman would sell cars in principe "higher max. engine revs = better/faster car". Difference is that average people tend to know more about cars than about computers: nobody would buy a car with shiny new 15 000RPM engine on rusty 20 years old chassis, without tires and seats as extra luxury option.

 

[Priit]

Sorry, double post.

 

[imgod2u]

<< PR=1.5*Frequency - 500. >>



The current PR system is based on the multiplier, each .5 increment in the multiplier brings about "100+" PR rating. Hence, AthlonXP "2000+" is a 12.5 multiplier, the "1900+" is 12 multiplier, the "1800+" is the 11.5 multiplier, the "2100+" is a 13 multiplier.



<< No it's not BS. Face the facts. AMD released the K6 in 1996, and was actually faster in Mhz than Intel at the time. Craig Barret then stated, "we will never let AMD get faster Mhz than us again". Intel did not want to be lagging in Mhz, so they purposely created a processor such. >>



AMD had a 2% market share in 1996, the implication that Intel even thought about AMD when planning their designs is worthy of a conspiracy theory. Only when the K7 came out in 1999 did AMD even begin to affect Intel's plans on anything (the CuMine's early release and the i815 chipset for example).

As far as the thing about memory, the 20-stage pipeline makes the P7 core extremely dependent on memory bandwidth (well, the higher MHz does). Processors (all single-core SISD processors) try to fetch one instruction every clock. Meaning a 2.4 GHz P4 will try to fetch 2.4 billion instructions every clock, a far cry from the 1.73 billion instructions per second fetched by the AthlonXP 1.73 GHz. You put a dual channel RDRAM solution and a 100MHz QDR FSB on the Athlon and I'll be you it wouldn't help it nearly as much (in fact, considering the increase in latency, it may actually hurt it) as the difference between a P4 with a 133MHz DDR FSB and memory jumped to a 100MHz QDR FSB and memory.

 

[Mats]

are you going to tell me that you believe that a tbird clocked equally to a p4 will perform about the same?


Hard to say who's side you're on here. Considering the 1.4 T-Bird is competition for a 2.0 Ghz P4, I would assume you are talking from an AMD perspective.

 

[boran]

what most ppls here tend to forget is that the average consumer has NO idea about computers, they only know more is better, here they sell P4 2,2 ghz systems with a TNT M64 (!!!) to keep the price about equal to an XP2000 with GF2 (okay I'm talking bout dell that might have something to do with it too)

now imagine same thing without PR:

the average consumer doesnt even touch an XP with a ten foot pole cous it is 500 mhz slower than the P4, amd gets virtually no consumer marketshare, only educated users like almost 99% of the ppls here on the board will even considder amd, but still half will go for P4, now how much market share would AMD have then, yeah u guessed it, about 2 %

their XP rating has kept their marketshare uphill from where they left it with the K7, I dont think intel made the P4 so scalable on purpose, it is just a decision that turned out that way.

the solution is a performance measuring system that is industry wide and defines the performance, this has to be done by an independent organisation tho, and HOW is also very difficult, will u measure it with games or so ? or only theorethical speed ...

I dont see this happening anytime soon, and if it happens it'll prolly be intel that will do it, because when we make the step to 64 bit there AMD has the advantage in mhz but not nececearely in performance.

 

[TSDible]

I guess if we really want it to be fair...

We would have to get an independent to come up with a new PR system. Then, ALL companies that manufacutured CPUs would be required to rate their chips based on that system. Now THAT would be interesting to see. Finally we would see how an XP relates to a PIV and a G4... etc. etc.

Intel would NEVER go for that though.

 

[imgod2u]

Or better yet, stop with the clockspeed fetish and post instructions per second (not instructions per clock, that's just about as useless as clockspeed). Apple has done this before with their "we can do a gigaflop".

 

[LikeLinus]

<< guess if we really want it to be fair...

We would have to get an independent to come up with a new PR system. Then, ALL companies that manufacutured CPUs would be required to rate their chips based on that system. Now THAT would be interesting to see. Finally we would see how an XP relates to a PIV and a G4... etc. etc.

Intel would NEVER go for that though. >>>

That's what i've been saying! No one seems to get that. We understand normal Joe Blow doesn't understand Mhz doesnt mean much right now (but it used to, so you really can't fault the customer - It's like telling them Vitamin D is good for them so they drink milk, then all the sudden it's bad, but they still drink milk out of habbit). But the problem is the Bias Nature of AMD's rating system, NOT the fact that they use one.

As stated serveral times. There needs to be an independant, non-bias, board of members put together who can create a set of benchmarks and scale for the performance of chips. This is just like the age old question of whos faster....Apple chips or PC chips. If there was a standard, all this would be solved.

Just cause consumers are uneducated....doesn't mean they can just magically make up their own new performance measure.

 

[Zukatah]

As long as it doesn't go out of hand like the PR rating of the Cyrix processor (yeah my father bought one of these for my birthday...) I don't see any problems with the PR rating. It is only giving AMD a chance to sell their products to the uneducated masses. Lots of people haven't even heard of AMD before and if they see a slower clock speed, they'll stay away from it for sure. In contrast, everyone knows Intel and I think confidence plays a big role in the purchase of a computer for someone that doesn't know much about it, they want to play safe.
Wrap up: PR rating for AMD is good since it gives them an opportunity to sell their products against similar performance products at a good price. If AMD's market share go up, eveyone will benifit because there will be price cuts from both side. So everyone should understand the logic behind AMD's move and agree that clock speed isn't eveything.

 

[HardWareXpert]

<<

<< Intel are also fooling people with the 400MhzFSB which is 4x100Mhz quadpumped, adding more pipes to the bus dont make things better, making the water flow faster down the pipes does. Very few apps and games take advantage of heavy bandwith thats why in anything using FPU the Athlon wiped the floor with P4. >>



You need to change your name and FAST!!

Adding more pipes? WTF? Where did you get that?


Are you saying that AMD is not fooling people when they claim their 133FSB is 266Mhz?

http://athlonxp.amd.com/technicalInformation/

AMD calls theirs 266Mhz because they transfer data 2 times per clock... once on the rising edge, and once on the falling edge. How is it that Intel is fooling people by saying that their is Quadpumped, while AMD is not about their double pumped? Intel's FSB tranfers data FOUR TIMES per clock cycle... two times on both the rising and falling edges. This is what gives you your 100x4, which is the same place AMD gets their 133x2.

Educate yourself...

AnAndTech Article

The bandwidth on RDRAM is so high, something like this capable of supplying that bandwidth to the CPU was needed. The 400Mhz (now 533Mhz) FSB of the P4 was needed.


At least you used the past tense "wiped". I have to give you credit for that. >>



Dude it's just a example (pipes) It's been proved that the P4's quad pumped bus and it's huge bandwith is pritty useless, 4x100Mhz FSB is no more faster than AMD's 2x133Mhz FSB, hardly any game or app takes advantage of huge bandwith.

AMD's coupled with DDR ram is 2x133=266Mhz FSB(more realistic), however try a explain people with SDRAM P4's and RDRAM for that matter are running at 400Mhz FSB, what BS, AMD know that huge bandwith figures mean nothing and in reality, P4 is 100Mhz FSB as the performance in real benchmarks prove it.

The artical says theroy(DDR, QDR FSB) and the benchmarks say otherwise and only Quake3 is P4 able to perform good over the Athlon because of it's high bandwith requirements.

Again AMD show on the links you provided that a 266MHz FSB in real world benchmarks works just as good as Intels QDR 400Mhz FSB.

 

[Mats]

...she cares about other things like a quiet system that doesn't heat up the room.

You are full of $#!t today my friend. As if a P4 system doesn't "heat up the room" as much. Please keep pathetic comments like this to yourself because no-one is buying your 12-year-old-mentality crap.

 

[LikeLinus]

<<It's been proved that the P4's quad pumped bus and it's huge bandwith is pritty useless>>

Show me where it's been proven, ya noob. I want this "proof".

 

[RedShirt]

<< It's been proved that the P4's quad pumped bus and it's huge bandwith is pritty useless >>



Yeah? Then why does the PIV's preformance go down the toilet when SDRAM is used?

And changing subjects, yes, That would be great if an indepentant organization would rate these chips, that would be fair.

My claim still stands that the XP rating system will have to be changed in the not to distant future... And I think this is a bad thing.

 

[Nemesis77]

<< My claim still stands that the XP rating system will have to be changed in the not to distant future... And I think this is a bad thing. >>



You still haven't replied to the points I made. Using MHz to determine performance would be equally deceptive as using PR-rating, maybe even more so. MHz is a good way to determine performance inside one CPU-family, but it fails horribly when comparing CPU's of different family. That's why we need new way to determine performance, and at least the PR-rating gives more true results than plain MHz does. And so what is the PR-rating is not the same as the MHz? People don't buy MHz, they buy performance.

 

[LikeLinus]

<<People don't buy MHz, they buy performance. >>

Apparently someone doesnt know Joe Blow Public. They buy based on the whole "bigger is better" theory.

 

[HardWareXpert]

<< Yeah? Then why does the PIV's preformance go down the toilet when SDRAM is used? >>




Because the P4 needs ram that can supply it with huge amounts of bandwith, RDRAM can but against your standard AthlonXP/SDRAM it's not very good at all. P4's performance is dreadful anyway so it needs RDRAM/DDR to make any kind of decent performance.

Fact is not many apps or game take advantage of RDRAM's huge bandwith, this is why AMD has the PR system to say number really mean nothing, however Intel are selling the numbers to everyone.

 

[Nemesis77]

<< <<People don't buy MHz, they buy performance. >>

Apparently someone doesnt know Joe Blow Public. They buy based on the whole "bigger is better" theory. >>



I'm aware of that. But it all boils down that they don't buy MHz, they buy performance. They just use MHz as a measure of performance. And like I have pointed out, that is flawed reasoning. By providing PR-rating, AMD gives another measure of performance, and that maesure has more basis in reality as plain MHz does.

 

[HardWareXpert]

<< <<People don't buy MHz, they buy performance. >>

Apparently someone doesnt know Joe Blow Public. They buy based on the whole "bigger is better" theory. >>



Yes, this is why AMD have the PR system on the CPU's, just think of it this way, P4 2Ghz but AMD's equivelent AthlonXP 2000+ gives you the same performance(more infact) but at 1.73Ghz(and cheaper), why the hell would you moan at that?

I dont understand you people at all, why the hell are you moaning when the performance of the AthlonXP is AWESOME anyway? for godsake!!!

 

[LikeLinus]

 

[LikeLinus]

Show me where it's been proven, ya noob. I want this "proof". >> I'm still waiting for proof.

<<RDRAM can but against your standard AthlonXP/SDRAM it's not very good at all.>> Ugh, silly noobs. Are you actually saying that a P4 w/ RDRAM is going to be slower and have less bandwidth than an XP w/ SDRAM?????

<<P4's performance is dreadful anyway so it needs RDRAM/DDR to make any kind of decent performance.>>

Please for the love of god quit talking out of your ass. I'd really like to see all your data to back this up.


FYI, it seems you are missing the entire point to this conversation (which i'm so not surprised). The problem isnt that they use a PR system. Its the fact they made it up on their own and it has no substance and is actually on a fluxuating scale and bases performance off OLDER AMD chips. This is VERY bias and has no real meaning in terms of performance chip vs chip (from other companies).

Please read what i wrote this time. Go back and read everything. You miss every point i make and whine about something else, or just totally ignore it.

 

[Rectalfier]

"AMD had a 2% market share in 1996, the implication that Intel even thought about AMD when planning their designs is worthy of a conspiracy theory. Only when the K7 came out in 1999 did AMD even begin to affect Intel's plans on anything (the CuMine's early release and the i815 chipset for example)."

AMD did only have 2% market share, but it rose drastically when the highly clocked K6 was released. Intel purposely engineered the Pentium 4 for high clock speeds, and in my opinion, they did it for marketing reasons. If you don't agree, that's your opinion and you're entitled to it, but don't call mine BS, or else I will argue with you untill I die.

P.S. I am not and AMD fanatic. I buy what is best at the time, performance and budget wise. That is why I just ordered a P4B533, P4 1.6A, and PC3000 DDR. AMD has dropped the ball, and will not regain it untill Hammer.

 

[HardWareXpert]

Hold on a minute, AMD made up the PR system, well yes of course they did, how the hell else are AMD supposed to tell people that a lower clock CPU is faster than a higher clock CPU of the competition.

AthlonXP 1800+

P4 1.8Ghz


The AthlonXP 1800+ in benchmmarks is able to out perform the P4 1.8Ghz, HOW SIMPLE DO YOU WANT IT?

AMD only went from 1.43Ghz(1600+) to 1.47Ghz(1700+) because that jump was equivalent to a T-Bird 100Mhz jump.

 

[RedShirt]

<<

<< My claim still stands that the XP rating system will have to be changed in the not to distant future... And I think this is a bad thing. >>



You still haven't replied to the points I made. Using MHz to determine performance would be equally deceptive as using PR-rating, maybe even more so. MHz is a good way to determine performance inside one CPU-family, but it fails horribly when comparing CPU's of different family. That's why we need new way to determine performance, and at least the PR-rating gives more true results than plain MHz does. And so what is the PR-rating is not the same as the MHz? People don't buy MHz, they buy performance. >>



Well, what about the Durons, they are sold as a Duron 1200 that is really 1200 mhz. Is AMD being deceptive is saying the actual speed of the Duron even though we all know it blows the hell out of a 1200 mhz Celeron? Maybe they are, maybe they aren't.

I am totally for an independant organization to come along and rate every processor on the market, by using an extensive amount of standard benchmarks and then giving a certain processor a rating..

For instance, maybe the would rate a 2.4 GHz Pentium 4, 1900 or something, and a 1.6 Ghz Athlon the same, who knows, this is just an example. If every processor were put through the same tests and a PR rating were given, I would value to rating much more than AMD just guessing numbers just in order for the uneducated public to think its a 2000 mhz processor or whatever.

Why doesn't AMD simply advertise that, clock for clock, their chips are better. They could pit a 1.6 Ghz Athlon vs a 1.6 Ghz PIV on a commercial and have two users, one frusterated with his PIV and one happy with his Athlon, we don't see this.

AMD once wanted to teach the public that MHz isn't everything, but then decided not to.

Are they being deceptive, YES. If the public thinks its a 2000 mhz processor, and it is not, THAT is deception.

 

[HardWareXpert]

<< <<It's been proved that the P4's quad pumped bus and it's huge bandwith is pritty useless>>

Show me where it's been proven, ya noob. I want this "proof". >>



HERE old P4 1.5Ghz/RDRAM againest a Athlon 1.2Ghz/RDRAM

Only in Quake3(engine) does P4/RDRAM perform better but since only a handful of games now use the quake3 engine that what you call pretty useless.

 

[RedShirt]

<<

<< <<It's been proved that the P4's quad pumped bus and it's huge bandwith is pritty useless>>

Show me where it's been proven, ya noob. I want this "proof". >>



HERE old P4 1.5Ghz/RDRAM againest a Athlon 1.2Ghz/RDRAM

Only in Quake3(engine) does P4/RDRAM perform better but since only a handful of games now use the quake3 engine that what you call pretty useless. >>



Well, this does, in fact show that the P4, or current applications, don't take advantage of all that bandwidth, or the bandwidith isn't necessary, but this will change the faster and faster the chips get.

As the PIV gets faster, all this bandwidth will become more and more useful. It obviously isn't useless even now or else the PIV's with SDRAM wouldn't have gotton so many bad reviews.