Intel disagrees with AMD's PR Rating

[ath0mps0]

Yes, AMD states, for legal reasons only, that the AXP model numbers are with respect to the Tbird. Everyone, including AMD, knows that it is really a comparison to the P4. AMD also left room in the initial comparisons to allow for the performance increase associated with the Northwood 13 nanometer core.

Some day the government (US in this case as both companies are US based) may have to step in and regulate a good performance measurement. Actually, now that I think of it, maybe it would be better for the EU to do this, as competition (meaning $$$$$) is the true king in the US. We can't even get a standard mobile telephone standard let alone anything else. A standard set of measurements that demonstrate performance at a CPU, memory, peripheral and total system level would be great. Then PC companies would be required to slap a sticker in plain sight on the side (or top, or wherever) of the box to indicate its performance - like the nutritional information box on food products in the US.

This type of regulation would prevent what Packard Bell used to do: They would stick a 133MHz labeled CPU in a box and sell it as such, even though the entire system was <b>under</b>clocked and maybe only running at 120MHz due to cheap components. Intel is currently doing about the same thing with their procs - except they leave the MHz up, just strip out some of the proc inards without telling the average consumer that the proc now performs more poorly than their previous procs (clock for clock - or even at significantly higher clocks). With at performance sticker on every PC you could take a quick look at the total system performance metrics and make a decision - this would keep vendors more honest and the proc manus wouldn't have control over the market with just MHz or model numbers.

I thought a thought, but the thought I thought wasn't the thought I thought I had thought.

 

[Intel_inside]

yeah I guesse intel misundertood the PR rating system, just like almost everyone else. But now I realize it's importance, I always need to know the relative performance of an XP cpu vs. that of a non-existant thunderbird model, doesn't everyone?

<i>My life wasn't complete untill I tried sse-2 optimized pong</i>

 

[dhlucke]

Just make up a new unit that multiplies all the core factors. Call it something cool and there you go. What are the core factors though? They have to be consistant and they have to last.

ie. Hz x FSB x IPC

<font color=red>God</font color=red> <font color=blue>Bless</font color=blue> <font color=red>America!</font color=red>

 

[amebapk]

true it's just the model name, doesn't mean it has to be the core speed.
intel mess up again

 

[ath0mps0]

Just make up a new unit that multiplies all the core factors. Call it something cool and there you go. What are the core factors though? They have to be consistant and they have to last.

ie. Hz x FSB x IPC

The core factors have to be meaningful. Hz, FSB and IPC are factors in performance, but aren't true measurements of performance.

Back to the car comparison:

RPM, Horsepower, Torque, axle ratios, cubic engine size, weight, tire size, gas tank size, etc. only tell an experienced person <i>approximately</i> what performance to expect in the vehicle; they don't help the average consumer.

Top speed, 0-60MPH/0-100KPH, quarter-mile/500M speed and time, braking distance, maximum G's in a turn, etc. are true total system performance measurements that a consumer can look at and see the equivalent performance of their purchase.

Why I suggest the EU get involved is to establish an international set of performance specs - these would probably have to change from year to year and vendors would have to update their specs to the current year and have them certified to sell any particular model. These specs would include every individual component as well as total system performance.

I thought a thought, but the thought I thought wasn't the thought I thought I had thought.

 

[dhlucke]

Why the EU?

<font color=red>God</font color=red> <font color=blue>Bless</font color=blue> <font color=red>America!</font color=red>

 

[ath0mps0]

The EU have been extremely involved in creating meaningful regulatory legislation (albeit radical in some cases), that spans an international community. The recent UK compliance with data privacy and protection is an example. US politicians would probably never do this as the industry lobbyists would chew them up and spit them out; in the end the performance specs would be meaningless and/or only be meaningful to a very small niche of consumers (like tax preparation or something like that).

Some might equate the EU to the UN, but this is now mostly inaccurate - the UN has no real power while the EU does - mostly because of money. Especially now that the euro is moving into circulation, the EU will start to exercise control over its member states (very much like the U.S. in its early stages). While EU based regulation wouldn't directly force the US and other countries into similar legislation, it would prove the possibility, create the infrastructure with the vendors, and give internationally connected consumers (directly or via the Internet) a good estimate of a similarly configured system from the same vendor.

I thought a thought, but the thought I thought wasn't the thought I thought I had thought.

 

[Crashman]

I earned it. I think Fredi added it to the list after another member of our forum suggested it. I don't know what the next one is, but it should be better.

What's the frequency, Kenneth?

 

[Matisaro]

yeah I guesse intel misundertood the PR rating system, just like almost everyone else. But now I realize it's importance, I always need to know the relative performance of an XP cpu vs. that of a non-existant thunderbird model, doesn't everyone?

Well, imagine when the 1500+ came out, now imagine no pr ratings, how would amd convince the buyer that a 1.33ghz axp is better than a 1.4ghz tbird, when it clearly is?

THe pr rating was needed, and its effects have been good for amd, it dosent matter how good the athlon is, if amd goes out of buisness selling it.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

 

[Atolsammeek]

I agree with you Ath0mps0s. But I think it would be the People at Intel who dont like it. They lose there wouldbe slow Mhz.

 

[scamtrOn]

you don't really buy that do you?

wouldn't you think that intel out of all would know what the PR rating is? think about it, its just another way of marketing by "stretching the truth". Intel might be behind AMD in terms of Mhz, but they are not stupid and i'm sure that that graph was not made only by one person and what i'm saying is maybe one person got it wrong, but not every one.


i think intel knows exactly what they are doing and are not confused. that is just another way intel tries to get computer hardware stupid people on their side to buy their poor performing/Mhz CPUs. intel does screw people as would any other company in intels situation.

good going intel, keep on screwing.

<A HREF="http://www.geocities.com/scamtron2000/Lochel.html" target="_new"> <font color=red>go to my site</font color=red> </A>

 

[scamtrOn]

who am i replying to?

um... i think intel_inside no?

<A HREF="http://www.geocities.com/scamtron2000/Lochel.html" target="_new"> <font color=red>go to my site</font color=red> </A>

 

[girish]

That line clears it up pretty well...based on tbird performance.

Page 8 of the official <A HREF="http://athlonxp.amd.com/includes/content/whitePapers/benchmarkingModelNumbering.pdf" target="_new">Model numbering doc</A> says:

All tests were run on the Microsoft® Windows XP operating system, as AMD
expects it will be relevant to most purchasers of x86 PCs over the next several years. The
results on the previous pages were obtained when the three suites of benchmarks and
applications were run following the aforementioned methodology. All results have been
normalized to a Pentium 4 processor running at 1.5GHz. Again, when viewing benchmark results and attempting to analyze their meaning, a normalization process is
useful. Detailed scores and individual results for all AMD Athlon XP processors can be
seen in Appendix B.

page 9 has all the relevent tables and explaination and figures in % improvement in performance over the P4 1.5 GHz. Clearly, this comparision is made with the Williamette model P4 and Palomino core AthlonXP. I am surprised as to how Intel defies this comparision on the basis of performance improvement offered by a later model of the P4! Guess AMD will have to re-evaluate the performance gain and scale it to Northwood(A) scores.

Then Intel will publich yet another document saying the results are not in line with the Northwood(B) line of their processors and them... AMD will have to restructure the figures again!

I still think the PR ratings are pretty good in view of Williamette Pentium4 they were marked against, but serve pretty well to compare to the NWs (at least the A versions) as well!

girish


<font color=red>Nothing is fool-proof. Fools are Ingenious!</font color=red>

 

[jlbigguy]

<font color=blue>"Considering that no other CPU manufactures fabricate a PR system to repesent their CPU performance I see little need for it. Also remeber that AMD and Cyrix did the PR system way back when. It failed too and look where Cyrix is now. The system is unessasary if AMD's CPU's are soo good place a commercial on at prime time and explain to america why. Telling half truths gets you nowhere fast."</font color=blue>

Telling half truths gets you nowhere fast. Spud, that is correct. Now, look at the half truths from Intel. The fastest processors. Measured by what? Clock speed? Isn't there a little deception there? Intel chose an architecture that allows a higher clock speed yet performs less instructions per second then the competitor. So while using the highest clock speed, performance is still on par with the competition.

Looks a little like Intel developed a product to sell to the consumers based on marketing hype (fastest clock speed), not by product performance. But that is the way business works.

If cars were sold by engine RPM, someone would create an engine that ran at 10,000. However, the transmission would only have 2 gears and the car would do 65mph at 10,000 rpm.

That is how the P4, Willamette or Northwood, appears to me.

There is nothing wrong with this, and if scaling the chip to 5 or 10ghz is possible for higher performance, then that is the way it is. But you cannot use the "half truth" of clock speed to try to explain away the competition that runs at a slower clock speed, offers similiar performance, and at a much lower price.

My opinion only......


<font color=blue>This is a Forum, not a playground. Treat it with Respect.</font color=blue>

 

[bum_jcrules]

No, I think that you did quite well. A lot of posters here are still in thier college years and won't take their higher level courses in corporate finance and accounting for a few years. Some are not even in high school yet.

I thought it was good.

<b>"Taurelilomea-tumbalemorna Tumbaletaurea Lomeanor" - Treebeard</b> :lol:

 

[bum_jcrules]

It is called "Full Disclosure."

If you fail to tell the client the positives AND the negatives you are misleading the consumer.

I have to do this with all of my clients when it comes to Investment, Insurance, etc. If I don't do it I can:
1. Be prosecuted
2. Serve jail time
3. Pay fines
4. Lose my corresponding licenses

It is misleading and wrong under the law. It is like telling someone that you can get at their principle back from an guaranteed fixed rate investment anytime but fail to mention to them that the is a 7% CDSC from the asset management firm. (CDSC – Contingent Deferred Sales Charge – Back Ended Fee) Hence when the client is expecting a check for $50,000, only gets $46,500 and then claims that I did not tell them that their was a fee if they took their money out too early. If I didn’t disclose that information, I would find myself with one or all of the four items listed above.

If Intel states that they have the best performance based on having the fastest clock speeds they are correct in their statement. However they are only telling half of the story. Their highest clock speeds are the fastest but they also have to say that the number of instructions per clock cycle is less than their last processor and thus the higher clock speed has a diminished return. (Comparing apples to apples instead of apples to oranges.)

It is illegal in medicine, in legal matters, in finance, and in every area of sales or servicing. If the rest of the corporate world is barred from doing it, Intel can't do it either. Use of misleading information is fraud.

<b>"Taurelilomea-tumbalemorna Tumbaletaurea Lomeanor" - Treebeard</b> :lol:

 

[Oni]

Guess AMD will have to re-evaluate the performance gain and scale it to Northwood(A) scores.

Why would they have to do this? their PR rating is not targeted at Pentium 4, so when Pentium 4 changes and gets a performance boost AMD doesn't have to do anything with their PR system. Its comparted to Tbird.
Supposedly with AMD PR system Athlon XP 1800+ gives the same performance as an 1800 MHz Thunderbird. Now we know that 1.4 GHz Tbird beats a 1.6 P4 in almost all benchmarks so do a little analyzing and you can see its easy to understand the PR rating isn't targeted AT the Pentium 4, but it does make people think twice before buying one.

"Why can't I be the man? I mean, I DO have harmony balls..." -epoth

 

[bum_jcrules]

This is to follow up AMD-Man's comment to Rick:

"VHS beat beta, didn't it?"

Yes they were incompatable but BETA was the better technology. It was smaller, had better magnectic surface for higher quality playback, cost less to produce, but was illegally forced out of the market by "DUMPING."

The foreign firms dumped thier product into the North American market and VHS was forced out of the market by an inferior and more expensive product. The foreign firms took a financial beating for discounting their product to be "Cheaper" in our marketplace while in their own it was 2-3 times as expensive. (Price fixing) Their product even had huge tarriffs placed upon it by the USA and they still priced it cheaper than Beta. Illegal marketing and sales practices kill competition.

<b>"Taurelilomea-tumbalemorna Tumbaletaurea Lomeanor" - Treebeard</b> :lol:

 

[ath0mps0]

Actually, Oni, the AXP model numbers <b>are</b> targeted at P4. AMD just stated that they were targeted at the Tbird for legal reasons. Without the artificially inflated - through removal of key circuitry - P4 GHz rates, AMD wouldn't have to do anything to combat Intel's GHz myth.

Now girish is also wrong; AMD doesn't have to re-evaluate for the NW(A) because AMD intentionally left enough room in the model numbers to account for the faster NW versions of the P4. Take a look at most benchmarks on the web and you will find that an AXP2000+ performs a little better overall than a P4 2.0 northwood and a lot better overall than a P4 2.0 willamette. It even exceeds the P4 2.2A on occasion.

I thought a thought, but the thought I thought wasn't the thought I thought I had thought.

 

[girish]

oh no! it seems nobody understood my sarcasm in my post!

I obviously dont expect AMD to re- and re-evaluate the PR ratings and update them with every new P4 core. I only tried to indicate that Intel's statement regarding the model numbering was a late afterthought, and that too with the wrong processor to compare with, stupid IMO.

ath0, I agree with you, the 1.67 GHz AthlonXP at times does beat the NW 2G2 so I guess I was right when I said the PR ratings were conservative enough for a larger range of P4s. In fact they could have rated 100 XP points for every 50 MHz instead of 66 MHz of real speed so that current XP2000 would have been really around the 2500+ mark!

Now, considering there are a lot of P4 models coming up, each with considerable improvement with various features like the FSB (400->533 with NWB, 533->667 with Prescott-A, 667->800 with Prescott-B) as well as cache upgrade with NW (and possibly later models sporting as much as 1 meg of L2) the number of variables certainly change so that having a single PR rating system will not be effective and accurate.

Later XPs (Tbreds and Bartons) can be no longer compared to Williamette P4 1.5 GHz since P4s at that time themselves will have some relative performance gain (or rating) over the Williamette! The PR rating is supposed to indicate some relation over the performance of contemprorary processors and not over a ancient model.

These PR ratings will get pretty confusing later on that it would actually be better to have a new base rating of the <i>current</i> (read contemprory) competition.

You certainly dont want a XP3500+ (w.r.t. P4 1.5 GHz) compared against a NWB 3 GHz which itself would be approx PR3500 (w.r.t. P4 1.5 GHz)! That would be real confusion!!

girish

<font color=red>Nothing is fool-proof. Fools are Ingenious!</font color=red>

 

[zengeos]

JC a bigger reason why VHS beat Beta is that VHS was cheaply licensed to companies for branding. SONY kept their licensing very tight. So, it became SONY against the rest of the industry.

Same basic issue with the IBM PS2 buses, etc compared to the industry. Remember...the more of something you sell and manufacture, the lower the cost becomes. VHS was being produced all over the place, so VHS makers were competing against each other for marketshare too.


Mark-

When all else fails, throw your computer out the window!!!

 

[zengeos]

"These PR ratings will get pretty confusing later on that it would actually be better to have a new base rating of the current (read contemprory) competition.

You certainly dont want a XP3500+ (w.r.t. P4 1.5 GHz) compared against a NWB 3 GHz which itself would be approx PR3500 (w.r.t. P4 1.5 GHz)! That would be real confusion!!"

AMD has stated that the PR rating system is only going to be used until an industry standard performance rating system can be developed for all CPUs, both Intel and AMD's.

When this will happen is anyones guess, BUT if it doesn't happen through industry self regulation it may possibly happen through government mandated action.

By the time several of these upgraded Pentium 4's come out Hammer should be nearing readiness and that is a whole new game.

Mark-

When all else fails, throw your computer out the window!!!

 

[girish]

I wish it would be that simple!

I know AMD, alongwith a number of other companies (excluding Intel I reckon) is working on Global Consumer Advisory Board (GCAB) which would standardise a set of performance benchmarks that would enable the average consumer to know the real performance of a system and help him choose his equipment. I hope these benchmarks take out the three basic factors 1. FSB, 2. Core speed and 3. Bitness (32 bit/64 bit) and offer a single unified performance rating which would actually indicate some sort of real world performance that the consumer is actually going to experience.

It should be just like evaluating how fast you reach a certain destination irrespective of what vehicle you use. Might sound strange or stupid, but it would concentrate on real miles/hr rather than gauging the BHP, tyre diameter, engine CC and miles/gallon for the vehicle!
The consumer can then decide for himself what miles/hr he can afford given a certain limits and parameters to his purchase.

girish

<font color=red>Nothing is fool-proof. Fools are Ingenious!</font color=red>

 

[girish]

Till the GCAB is in place, I guess this PR system will stay. Hope its out by the time next P4 arrives, making PR more and more confusing!

<font color=red>Nothing is fool-proof. Fools are Ingenious!</font color=red>

 

[Tiberius13]

For a company that is complaining about PR (Intel), they certainly don't hesitate to spit out PR themselves.

I think all of us are completely aware of the fact that the business world revolves around marketing (not necessarily the best product). That is why Microsoft rules the Operating System and Office suite market... or why Coke or McDonalds are international powerhouse brands, etc. Marketing is incredibly important to a companies success. Naturally this means that companies will not just sit back and accept another companies 'marketing' but will instead counter with PR of their own.

In this case, Intel claims that AMD is putting out misleading PR... but then they present graphs that are misleading in themselves.

1) They show the gap between actual Ghz and the AMD PR rating getting larger - but they fail to mention that this makes perfect sense as the performance gap would not be a static amount (such as .1 Ghz), but instead would be a % performance gain (ie. 20% faster). Therefore, the gap would increase in Ghz as the speed increases - but remains roughly constant as a percentage of performance (duh to Intel on this one... thanks for the "PR" Intel... shame on you for trying to fool people who don't understand the issue).

2) Intel then shows that an Intel processor outperforms the equivalent AMD PR rated processor in certain tests. This obviously only shows tests where they were faster - it does not show a balanced view as it omits any tests where the AMD processor may have been faster (nice PR!).

Anyways... just my thoughts... Intel may indeed be faster at 2 Ghz... but their PR isn't the place I would look to determine that. Third party reviews are always the best place to look for such information.

But... congrats on Intel for continuing the PR battle! It just makes things more interesting...