PR Mobile <> PR Desktop ?

Kemche

Distinguished
Oct 5, 2001
284
0
18,780
PR Rating has been discussed many many times and everyone would agree that AMD is being conservative on it's side. But how about Mobile version it's processor. I think AMD is just mis-using this PR rating for Mobile. Here's the rating for newley released mobile processors...

Athlon 4 1600+ = 1400Mhz
Athlon 4 1500+ = 1300Mhz
Athlon 4 1.2Ghz = 1200 Mhz

Ratings for Desktop Processors

XP 2100+ = 1733 Mhz
XP 2000+ = 1667 Mhz
XP 1900+ = 1600 Mhz
XP 1800+ = 1533 Mhz
XP 1700+ = 1467 Mhz
XP 1600+ = 1400 Mhz
XP 1500+ = 1333 Mhz

All of these processors are based on Palomino core. Also the mobile processors have 200Mhz FSB instead of Desktop's 266Mhz. So you should also agree that Preformance will be different between Mobile and Desktop with the same Mhz processor. So my question is how come PR 1500+ on desktop is 1.333Ghz and it's 1.3Ghz for Mobile. And the difference between 100Mhz on Mobile equal to 100PR. But on desktop it's only 66.67 Mhz equal 100PR. Is AMD mis-using this PR rating for mobile processors??????

KG


<b>"Hey! It compiles! Ship it!"</b>
 

AMD_Man

Splendid
Jul 3, 2001
7,376
2
25,780
I agree with you in this case, Raystonn. AMD is misusing their own PR rating.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
It has happened every time a PR rating has ever been used. It was only a matter of time before it would happen again.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
Condering their auditor for PR ratings is Arthur Andersen, it does not surprise me. We need a new way to indicate performance.

A better way to indicate performance might be to list the clockspeed and number of pipeline stages for the processor, along with the clockspeed and width of the FSB. Something like: Pentium 4 2.4GHz/20/400/64. This would indicate a Pentium 4 running at 2.4GHz with a 20 stage pipeline, and a 400MHz 64-bit FSB. I believe this would cover the important performance aspects of all processors, would it not? An Athlon 2100+ would be: Athlon XP 1.73GHz/10/266/64.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 

AMD_Man

Splendid
Jul 3, 2001
7,376
2
25,780
A better way to indicate performance might be to list the clockspeed and number of pipeline stages for the processor, along with the clockspeed and width of the FSB. Something like: Pentium 4 2.4GHz/20/400/64. This would indicate a Pentium 4 running at 2.4GHz with a 20 stage pipeline, and a 400MHz 64-bit FSB. I believe this would cover the important performance aspects of all processors, would it not? An Athlon 2100+ would be: Athlon XP 1.73GHz/10/266/64.
Good idea! If only both Intel and AMD can agree to do that!

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
This would also simplify indicating what you have done with an overclock. For example, a processor I am currently burning in (a 1.6A) is currently at 2.407GHz/20/600/64. You can easily see what I did using these simple numbers, and you can easily see what kind of processer I am talking about by looking at the number of pipelines and FSB width.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 

FatBurger

Illustrious
Only problem being, that would confuse the average consumer. They're confused enough as it is.

And yes, I knew Arthur Andersen was AMD's auditor.

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft
 

AMD_Man

Splendid
Jul 3, 2001
7,376
2
25,780
Take this
2.407GHz/20/600/64
And tell people the higher the first number the better. The lower the second number the better. The higher the second number the beter and higher the last number the better.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
The average consumer should not be making purchases without help. The numbers exist to give the salesman something to point at when he explains that X is faster than Y. Would you rather a salesman point at the PR rating and say "this number is larger, therefore it is faster"? Using something like what I suggested would break down performance into a few numbers similar to how it is done for hard drives. The salesman can simply explain that higher is better for all of the numbers except the second, for which lower is better.

As compared to a hard drive, the CPU clockspeed would be similar to the rotation speed (7200 vs 5400, etc.), the number of stages in the CPU's pipeline would be similar to the seek time, the FSB clockspeed would be equal to the controller bandwidth (i.e. ATA100, ATA133, etc.), and the FSB width would be equal to the width of the controller's cables to the hard drive.

None of this would actually need to be explained to the average consumer though. The salesman would simply point out that bigger is better except for the second number, for which smaller is better.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 

FatBurger

Illustrious
You mentioned hard drives with similar ratings, but people don't use them all as they should. Not even people who know roughly what they mean.

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
Well, salesmen are going to sell whatever crap they have, regardless of performance. Two things you simply cannot trust are salesmen and PR ratings. If you want the facts then you talk to someone knowledgable. Someone who is knowledgable would benefit from these extra numbers when comparing the performance of processors and explaining it to the newbie.

This is going to end up turning into something similar to the car market. We are going to end up with model numbers and such just to make things 'sound cool.' "Introducing the brand new Athlon 10000 Special Edition! The Athlon 10000 SE comes with a shiny new IHS/spoiler to make all your friends gasp with delight!"...

If the salesmen want a spiffy name, let them have it. But it should be prepended to a set of numbers similar to what I gave. Introducing the Athlon 10000 Special Edition 2.0GHz/10/266/64!

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 

Matisaro

Splendid
Mar 23, 2001
6,737
0
25,780
All of these processors are based on Palomino core. Also the mobile processors have 200Mhz FSB instead of Desktop's 266Mhz. So you should also agree that Preformance will be different between Mobile and Desktop with the same Mhz processor. So my question is how come PR 1500+ on desktop is 1.333Ghz and it's 1.3Ghz for Mobile. And the difference between 100Mhz on Mobile equal to 100PR. But on desktop it's only 66.67 Mhz equal 100PR. Is AMD mis-using this PR rating for mobile processors??????

Hmmm, dont the mobile processors run on a sdram platform, and would that not lower performance.


It appears to me that amd is lowering the pr rating of its mobile parts due to the fact the benchmark lower than their desktop counterparts on the average system.

But I agree its a tad weird.

"The Cash Left In My Pocket,The BEST Benchmark"
 

SammyBoy

Distinguished
Mar 16, 2001
689
0
18,980
Problem being, though, is that when it comes to salesmen at Best Buy, Circuit City, etc., they are nothing more than high school students or people who need a second job to suppliment their income. They are just a bit more educated than Joe Consumer. So, most often, you get two people ignorant about the finer points of a computer, who both just look for bigger numbers to indicate better performance. In that case, Intel would win, always, since their numbers are bigger. Wouldn't a better method than yours Ray be IPC times clockspeed? So, a AXP 2100+ would have a rating of 15597 and a 2.2GHz P4 would be 13200. Granted, it shows AMD to be in the lead, by quite a margin. So, there is no answer, now is there. Consumers and salesmen think big = better. Right now, things are not in sync with anything, so f--- it... let the consumers keep making uninformed decisions. As long as they keep buying, the prices stay low.

Pardon the cynicism, but I'm tired of trying to educate people, 'cause they don't want to be educated. Let AMD and Intel worry about it, as it only affects their high-end anyway. So, I give up... let the ignorant remain ignorant. What do we care anyway? We buy based on performance and price, and sometimes overclockability... I'd buy a 500MHz processor if it whipped the 1.5GHz and 2.0GHz competetors (as long as it's PC, and not Mac)

-SammyBoy
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
... both just look for bigger numbers to indicate better performance. In that case, Intel would win, always, since their numbers are bigger.
Well, we could always change the second number into the inverse of the number of stages in the pipeline, perhaps multiplied by 100. Then bigger would always be better. Athlons would use 10.0 and the Pentium 4 would use 5.0.


Wouldn't a better method than yours Ray be IPC times clockspeed?
IPC is not a measurable quantity. Every instruction takes a different number of clocks to execute and every application uses a different selection of instructions. This would always take us back to the question of which applications to measure. I prefer an inverse of the number of stages of the pipeline if you really demand that all numbers mean that bigger is better.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 

zengeos

Distinguished
Jul 3, 2001
921
0
18,980
AMD didn't produce a 1.3ghz Palomino for the desktop. Their lowest desktop stepping was 1.33ghz.

The closest PR rating to the desktop equivalent is PR 1500.
The next would be PR 1600 so the 100mhz jump there. It looks like AMD chose not to release a 66mhz stepping and did a full 100mhz stepping instead is all.

Mark-



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

AMD_Man

Splendid
Jul 3, 2001
7,376
2
25,780
...but a 50MHz jump is. An Athlon 4 1500+ running at 1.35GHz would be more reasonable.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
 

Schmide

Distinguished
Aug 2, 2001
1,442
0
19,280
They could just drop the “+” from the mobile processors, I doubt they would accept a “-“.

All errors are undocumented features waiting to be discovered.
 

eden

Champion
Maybe Mat is somewhat right about the relation in FSB.
Maybe the notebook versions have a different FSB or such? In any case I think we should try to contact AMD on it, or look on their website.

Ray this isn't the beggining of PR madness, AMD might probably change the Hammer PR relation to AXP, and if it does, then the Hammer has serious strengh then. But just note that this is a very variable system, like MHZ and FSB versions of one processor.

--
For the first time, Hookers are hooked on Phonics!!
 

ath0mps0

Distinguished
Feb 16, 2002
579
0
18,980
IPC is not a measurable quantity. Every instruction takes a different number of clocks to execute and every application uses a different selection of instructions. This would always take us back to the question of which applications to measure. I prefer an inverse of the number of stages of the pipeline if you really demand that all numbers mean that bigger is better.
The problem with all of these approaches is that they don't actually measure total system performance. Some of the benchmarks attempt to do this, but some are sponsored by the vendors and have enhancements for one brand but not another and therefore can't really be trusted. In addition you always get back to the question: which applications are important to the user?

The only true solution to this problem would be for an industry association or the governments to develop independant ratings for individual components and total system performance. The processor ratings would need to be similar to horsepower and torque in the auto industry - something that can be independently tested and verified - right now clock speed is just maximum RPM ratings. Total system performance ratings would have to be similar to max MPH/KPH and 1/4 mile time/speed ratings. This is something that AMD has been trying to do, but Intel refuses to participate because their bread is buttered with the GHz clock speed ratings.

I don't see true cross vendor performance ratings ever happening until (and if) AMD's procs outperform Intel's to the point that Intel can no longer charge a premium based on name alone. Whether or not this will ever happen, I don't know, but AMD is well on their way with their small die/high IPC/x86-forever strategy.

Don't get me wrong, Intel makes a quality part, but they have until just recently (past 2 years) been able to charge whatever they please for their procs. Since the release of the Athlon, Intel has had to scramble to keep ahead on the MHz curve (i.e. the1GHz PIII release and the 1.13GHz PIII recall). Intel, intentionally IMHO, threw a wrench in the works by introducing a higher-clocking lower-IPC processor with the P4. AMD simply countered by introducing the model rating. The validity of the ratings - although certified by Arthur Anderson - are being determined by the market.

Until comparable laptops with P4Ms and A4Ms are released (something Intel has used its "preferred vendor" status to prevent up until now), we will never know. Just try to buy a corporate grade desktop replacement laptop (high performance, three spindles including DVD/CDRW combo, large 15" screen, etc.) with an Athlon proc - I haven't found one yet. Only the Sony Viao FXA49 even comes close - but only at 1.2GHz. HP only offers the Athlon 1.2GHz with a 12" screen and only the 1GHz comes with a 14" screen, both on their consumer line. Even some of HP's Celeron models get 15" screens. HP used to offer a 15" screen with the Athlon, but Intel successfully pressured them to eliminate it. Compaq is a little better, actually touting the business model NV115 1500+ as coming with "a large 14 inch screen," and a custom built 1600+, but offers no AMD model with a 15" screen.

Intel has already shown us the way to create a concensus in the industry - anytime that Intel becomes the performance/technology/recall underdog/pariah, everyone else (relatively speaking) aligns against them. :tongue:

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

SammyBoy

Distinguished
Mar 16, 2001
689
0
18,980
Well, I'm just saying that the assumption of salesmen being intelligent about the product they are selling is false. <i>Some</i> are top-notch, and take their job seriously, but most only have a passing knowledge of computers. I'm just saying that we shouldn't have to worry about this, it's AMD and Intel's problem. Who cares what they start using for a speed rating. In the end, all that matters is that <i>we</i> have the knowledge and use it correctly. They could call their processors "S--- on a chip," and as long as they performed well and had a reasonable cost, we'd still buy them.

-SammyBoy