Sign in with
Sign up | Sign in
Your question

Miffed At AMD

Last response: in CPUs
Share
May 13, 2003 8:48:00 PM

Okay, I know what you're probably all thinking. Here goes another anti-AMD rant, right? Well let me start off by saying this before you turn the bias-filtering on:

I'm impressed that the AXP3200+ beat the AXP3000+. It shows that there is some life left in the AXP. I'm also completely unsurprised by the AXP3200+. Actually, that's not entirely true. I had predicted that AMD would just up the FSB to 400 and keep the clock the same or less than the AXP 3000+ to reach the 3200 label. The fact that it is a tiny bit faster instead of the same speed or slower did come as a mild surprise, a positive surprise. :) 

And if AMD wants to hold onto tne pretty insubstantial claim that their rating system is in comparison to the T-Bird, then fine. I don't believe it for a second, but I'm not going to get upset over it.

What gets my goat is that now AMD is pushing reviewers to use custom software, undoubtedly <i>optimized</i> software, for the reviews. Not only that, but they're also pushing reviewers to use <i>very</i> specific hardware and firmware settings to bias the benchmarks even further.

How many review sites will do just what AMD tells the reviewers to do? How many review sites will bias their results just exactly as AMD tells them to? How many review sites will never even mention this biasing?

AMD can make up ludicrous PR numbers all they want. I don't care. They can call the next CPU 'Bob' for all I care. So long as I have fair benchmarks to determine the true value of a CPU, I don't give a flaming fairy fart what the rating number or what the MHz is. I care about the performance that I will get for my money. That's it.

But when AMD starts pushing reviewers to bias those benchmarks, when AMD screws with my ability to research the actual performance that I should be able to expect from my hardware, <i>that's</i> when they've gone too far. AMD has most definately crossed the line now. It's one thing to mislead customers about the actual performance of the chip based on it's name. After all, it's just a name and anyone can compensate for that easily enough with just a little bit of research. It's something else entirely to get reviewers to bias their benchmarks so that customers can't even make an educated judgement anymore.

AMD has lost major kudo points in my book. And a big thanks to THG for pointing out this new and disgusting tactic from AMD.

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>

More about : miffed amd

May 13, 2003 9:27:20 PM

Before you getted "miffed" about AMD wanting reviewers to use optimized software, keep in mind that quite a bit of the software used in the THG tests are optimized for the P4. True, these programs are written out of the box that way, but that doesn't mean AMD is going past the line by offering thier own optimized software. AMD really needs to work with the software developers and have them write programs optimized for their processors. But since that hasn't happened, and the P4 looks better on the tests because the progams are optimized for it, I don't think it is such a bad thing for AMD to offer their own optimizations.
May 13, 2003 11:00:36 PM

To keep apace with the upcoming P4 3.2, AMD would've needed to clock the XP3200+ at 2.5 Ghz, no less. But as the other Bartons were clocked lower, I guess this release bears a tiny positive mark. But it's just tiny, since at 2.2 GHz, the 3200+ would be slaughtered by the P4 3.2.

While the other guy who replied to this mail pointed out that Intel has SSE2 programs, that definitely puts the P4 in a greater light than it would have been in otherwise, you're absolutely correct. AMD's way of pursuing this program preference is a worse way of doing it, to purposely and unfairly pull extra performance to themselves. SSE2 is more like a built in feature of the P4, while this act of AMD is nothing more than a desperate way to make themselves look better, regardless of the cost.

AMD is losing lots of credibility in my "book" as well. I remember you as stubbornigly trying to believe I am an AMD fanboy with nothing open-minded to say. If I remember correctly, last time I had a discussion with you, you didn't answer my last post. If you remember my name, I hope we could finalize this discussion in a creative and open-minded way for both of us.


My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
Related resources
Can't find your answer ? Ask !
May 13, 2003 11:11:42 PM

Quote:
Before you getted "miffed" about AMD wanting reviewers to use optimized software, keep in mind that quite a bit of the software used in the THG tests are optimized for the P4. True, these programs are written out of the box that way, but that doesn't mean AMD is going past the line by offering thier own optimized software. AMD really needs to work with the software developers and have them write programs optimized for their processors. But since that hasn't happened, and the P4 looks better on the tests because the progams are optimized for it, I don't think it is such a bad thing for AMD to offer their own optimizations.


I don't see the logic in that at all. The whole reason behind benchmarks is to tell you, the consumer, what you can expect "out-of-the-box", if the software is optimized for the P4 out-of-the-box, then that's the kind of performance consumers will get and the review sites showing it is giving them information they want. There's absolutely nothing wrong with that. It's the review site passing on information.
Now, when you have a review site *change* their benchmark to something *different* than what the consumer will get, then it's inaccurate and misleading information. The consumer looks at the review, buys the product, runs the software and doesn't get the performance he expected. *That* is misleading. It's all about informing the consumers with the results they can get. The sites that do that accurately are doing their jobs, the sites that don't aren't. And it's AMD's fault for trying to convince sites to focus more on making them look good than to provide the consumer with what they should expect.
Of course, Intel does this too. But there's a very negative and scrupulous watch over this (AMD's bitchfest about the Sysmark thing) and so, very little of it gets by. But no one seems to mind it when specialized software is used that puts the Athlon in a good light. It's a double-standard and one that I'm not surprised comming from the fanboys.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
May 14, 2003 3:10:03 AM

Quote:
I don't see the logic in that at all. The whole reason behind benchmarks is to tell you, the consumer, what you can expect "out-of-the-box", if the software is optimized for the P4 out-of-the-box, then that's the kind of performance consumers will get and the review sites showing it is giving them information they want. There's absolutely nothing wrong with that. It's the review site passing on information.

Couldn´t agree more! What if most software is written with SSE2 optimizations? That´s an out-of-the-box advantage for the P4! Why do people fail to recognize that?...
May 14, 2003 4:49:11 AM

couldnt agree more, though personally i still like AMD's chips, they perform well for the money, especially if you primarily play games, i just get tired of inflated pr ratings, call it what it is, so what if you cant lauch the fastest processor, the majority of money is made in the low end midrange performance sector

the funny thing is, as thg showed, the athlon can easily be clocked too 2.5 ghz with regular cooling, i wonder why AMD wont simply bump up the speed significantly and regain some consumer confidence

"Going to war without France is like going hunting without an accordion."
- Gen. (Ret) Norman Schwartzkopf
May 14, 2003 12:38:28 PM

Quote:
Now, when you have a review site *change* their benchmark to something *different* than what the consumer will get, then it's inaccurate and misleading information. The consumer looks at the review, buys the product, runs the software and doesn't get the performance he expected. *That* is misleading.

Exactly. Thank you imgod2u. I couldn't have said it better myself.

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>
May 14, 2003 12:46:46 PM

Quote:
Before you getted "miffed" about AMD wanting reviewers to use optimized software, keep in mind that quite a bit of the software used in the THG tests are optimized for the P4. True, these programs are written out of the box that way, but that doesn't mean AMD is going past the line by offering thier own optimized software. AMD really needs to work with the software developers and have them write programs optimized for their processors. But since that hasn't happened, and the P4 looks better on the tests because the progams are optimized for it, I don't think it is such a bad thing for AMD to offer their own optimizations.

Sorry jardows, but your counterpoint just doesn't hold any water.

Yeah, the P4 has SSE2 that the AXP doesn't have. Yeah the software <i>may</i> be optimized for the P4. (<sarcasm><i>Like it isn't also optimized for the Athlon in any way, shape, or form.</i></sarcasm>) But that's exactly the software that the consumer is getting. That's the exact performance that the consumer can expect from the software. So there's no dishonesty involved so long as the software that is benchmarked is exactly the same as the software that is sold to the consumer.

What AMD is doing is completely different. They are telling the reviewers to use specially-optimized software for the benchmarks that the consumer will never be able to purchase. The consumer will never see those benefits because the software that they buy will be different.

If AMD wants optimizations in the benchmarks, they should work <i>with</i> the software developers to change the software and the benchmarks. They should <i>not</i> be providing their own 'special version' dlls.

jardows, maybe it isn't fair that software is often optimized for the P4 before it is optimized for the Athlon. Then again, look at the market share difference between AMD and Intel. Look at the tools that Intel both sells and provides for software engineers. (Yes, Intel provides many <i>free</i> software libraries besides their special compilers.)

Frankly, the reason that software isn't as optimized for the Athlon as it is for the P4 is simple and only two part. Part one: AMD doesn't go to nearly the same lengths and doesn't provide nearly as much assistance to software developers to optimize for their CPU. They just don't provide nearly the same amount of resournces to the software development community as Intel does.

Part two: The market share is still heavily in Intel's favor. <i>Most</i> customers are going to be Intel customers. As a <i>business decision</i> the software developers cover the largest customer base that they can first and work down from there. It's the only sensible way to do it. And since Intel is their largest customer base, Intel get's first dibs on the optimization considerations.

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>
May 14, 2003 1:07:44 PM

SSE2 should not even come into the picture. The bottom line is do both processors run the same, out of the box code at the same speed? As a consumer, I don't care if the code is opimized or not. The product should deliver what the manufacturer promises.

AMD has done a commendable job keeping the Athlon XP competitive with the P4. No longer equal, but still competitive. The shame is that their top of the line chips are about the same cost as the P4, and not $100 (or more) less expensive (when compared to Intel) as they were 2 - 3 years ago.

The mid range processors (about 2400) AMD's PR rating is more realistic, and have a great price/performance advantage over Intel. But the top of the line performance is owned by Intel.

<font color=blue>This is a Forum, not a playground. Treat it with Respect.</font color=blue>
May 14, 2003 1:26:22 PM

Why is it such a big deal that you can get better perofrmance out of your athlon without changing the way the P4 does things? If you were the comsumer running these apps, why the hell should you care if changing a dll gives you better performance?
Does it do anything to the P4 score? No.
Does it give you better performance by cutting corners? No.
Getting more performance out of your processor by switching a dll shouldnt be an ethical issue like many of you are making it out to be, you really need to get off the computer if you take useless benchmarks that seriously. And Tom needs to stop taking it so [-peep-] seriously too. Instead of turning off APIC and getting better performance with no drawbacks, he just makes some snide comment about how he cant figure out how it gives you better performance and doesnt even TEST IT, what an idiot. Other review sites actually tested it and it gave much better performance in desktop apps without causing any hardware problems
May 14, 2003 1:37:45 PM

Quote:
To keep apace with the upcoming P4 3.2, AMD would've needed to clock the XP3200+ at 2.5 Ghz, no less. But as the other Bartons were clocked lower, I guess this release bears a tiny positive mark. But it's just tiny, since at 2.2 GHz, the 3200+ would be slaughtered by the P4 3.2.

Everyone keeps saying that the AXP3200+ should have been 2.4-2.5GHz or should have been an AXP 2900+. I don't entirely disagree from the perspective of my desires, but I am also completely unsurprised that AMD didn't do this from a business perspective.

Frankly, the 2.4GHz AXP3200+ would have been a ludicrous thing for AMD to do for a few reasons:
1) It would leave them with <i>no</i> further headroom at all to try and counter Intel with later. (Because that pretty much is <i>the</i> limit of the current Athlon die revision and I don't see AMD bothering with the Athlon anymore until they get their .09micron process down.)
2) The yields on 2.4GHz (or higher) AXPs that could actually be sold as such (and not as special overclocked chips) would probably be so bad that they'd have hardly any per batch. Most AXPs won't get there without serious cooling <i>and</i> a core voltage increase. The percentage of AXPs that can hit 2.4GHz with their retail cooling and stock voltage is miniscule. We like to rag on Intel's latest heat-monger CPUs, but the AXPs at 2.4GHz and higher would be putting out just as much (or really more) heat, but they have a significantly smaller amount of surface area on their die to transfer that heat to the heat sink. The dies are just simply too small. (And rectangular at that, which kills surface area. Square would be ideal.)
3) AMD's PR rating is already set on a certain path. Now, I may not fully agree with that path, but that's not really the point. The point is that for AMD to suddenly jump the processor speed just for a 200 rating increase would really screw up what little shred of sanity is left to their rating system. Just as for them to lower the ratings on any of their CPUs at this point would severely screw that up as well. AMD laid their ratings path out and now they have to follow it.

So honestly, I don't have any quarrels with AMD over the PR/MHz that they gave the chip. I may not agree with their PR scheme in general, but the chip fell perfectly into their scheme with much less confusion than some of their others, so I can't fault them for that at all.

Quote:
I remember you as stubbornigly trying to believe I am an AMD fanboy with nothing open-minded to say.

Not likely. If anything I made an <i>observation</i> based on your posts. There is a considerable difference between observing and believing. Observations change whenever that which is being observed changes. Beliefs change only when our own personal judgements change. There is only one person in all of THGC that has been so consistant and memorable that I hold any beliefs on, judgements of. Everyone else I just make observations of, which hold no judgement and therefore change anytime the posts change.

Quote:
If I remember correctly, last time I had a discussion with you, you didn't answer my last post.

That's entirely possible. Like any imperfect situation, I cannot be here 24/7/52<font color=blue><b>*</b></font color=blue> and as such some conversations become lost to me. Such is real life. I do not pretend to make the pretense that I will always be here to continue every conversation. Frankly, I just don't have that kind of time and as such I have to prioritize what I have.

Quote:
If you remember my name, I hope we could finalize this discussion in a creative and open-minded way for both of us.

When I am not self-animated, I am merely the mirror which through you may see yourself. If you are creative and open-minded, my replies will be creative and open-minded. If you are not, I am not. It is simply a matter of giving back the same respect given, which is exactly my modus operandi.

I have to ask though, which discussion are you referring to? This one or whichever one that I have not continued in the past?



-----footnote-----
<font color=blue><b>*</b></font color=blue> = It is a common <b>mis</b>label for people to state 24/7/365. This however litterally would mean 24 hours in a day, 7 days in a week, 365 weeks in a year. That is, of course, insane. The proper way would either be 24/365 for 24 hours in a day, 365 days in a year, or 24/7/52 for 24 hours in a day, 7 days in a week, 52 weeks in a year. I only make this point because I am campaigning to raise the awareness of this common mistake and bring it up any time that I use 24/7/52.

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>
May 14, 2003 1:56:58 PM

Quote:
SSE2 should not even come into the picture. The bottom line is do both processors run the same, out of the box code at the same speed? As a consumer, I don't care if the code is opimized or not. The product should deliver what the manufacturer promises.

Do you realize just how contradictory you are in just that one paragraph?

A processor is nothing more than a finely crafted lump of silicon and metal. By itself it is worthless. Give it electricity and it does nothing. It doesn't matter at what frequency the internal clock runs at, because nothing would happen. It doesn't matter how much L2 cache it has, because nothing would change its state.

To make a processor do <i>anything</i> you need to feed it instructions. You need to give it code. You need to give it data. When we benchmark, we are not testing the CPU's ability to jump x number of feet. We are not testing the CPU's ability to run a 100 meter dash. We are not testing the speed of any internal timing device. We are not testing the size of it's internal storage capacity.

What we <i>are</i> testing is that CPU's ability to run code, to process data, to execute instructions. What code/data/instructions exactly? Why Executable X, Y, and Z. How well does this chunk of silicon process the code of Executable X? How well does that chunk of silicon process the code of Executable Z?

<i>THAT</i> is "what the manufacturer promises". The manufacturer promises that this chunk of silicon runs Executable X this much faster than that chunk of silicon<font color=blue><b>*</b></font color=blue>. It doesn't matter <i>how</i> any chunk of silicon executes that code. It doesn't matter what means a manufacturer implemented in their hunk of silicon to ensure that Executable X runs yea fast. What matters is quite simply just how well this hunk of silicon runs the code given to it compared to that hunk of silicon running the exact same executable.

-----footnote-----
<font color=blue><b>*</b></font color=blue> = Other promises are that this chunk of silicon uses this much less electricity and generates this much less heat than that hunk of silicon, and that this chunk of silicon costs this much less than that chunk of silicon.

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>
May 14, 2003 2:14:32 PM

Quote:
Why is it such a big deal that you can get better perofrmance out of your athlon without changing the way the P4 does things? If you were the comsumer running these apps, why the hell should you care if changing a dll gives you better performance?
Does it do anything to the P4 score? No.
Does it give you better performance by cutting corners? No.
Getting more performance out of your processor by switching a dll shouldnt be an ethical issue like many of you are making it out to be, you really need to get off the computer if you take useless benchmarks that seriously.

The big deal is that these special dlls are <b>NOT</b> available to the consumer. The results from the benchmarks do <b>NOT</b> represent the same results that a consumer would get with the software that they buy off of the shelf. <i>That</i> is why it is a big ethical issue. <b>NO</b> consumer will <i>ever</i> see the same results that the reviewers are seeing with these special dlls and system tweaks.

Quote:
you really need to get off the computer if you take useless benchmarks that seriously

If you enjoy being manipulated and lied to, fine. That's your bag. Enjoy. Me, I don't enjoy that sort of thing. I have this nasty penchance for desiring the truth, the whole truth, and nothing but the truth. Call me crazy, but that's just my bag.

Indeed, morals, ethics, these are issues that are unique to each individual and again to each social structure. Perhaps in your ethics this is an unimportant issue. Fine. Then don't care about it. I won't make you. But <i>I do care</i>, and I <i>will</i> make an issue of it. If stands aren't taken over the little things, then just how far can these issues be pushed before the same people start making big issues? It is easier to stop it while it is a small thing than to let it grow and get worse and fight it then.

Quote:
And Tom needs to stop taking it so [-peep-] seriously too. Instead of turning off APIC and getting better performance with no drawbacks

<b>NO</b> drawbacks? Please do explain how it has absolutely <b>no</b> drawbacks when it limits the expandability of the entire PC.

And do please explain how limiting the expandability of the platform in a way that virtually <i>no one</i> would <i>ever</i> do is not biasing the results of the review away from the results that the typical consumer would get.

Quote:
Other review sites actually tested it and it gave much better performance in desktop apps without causing any hardware problems

Other than that they couldn't actually add any hardware to their PC anymore you mean. <sarcasm><i>Because you know that no one out there <i>ever</i> adds PCI cards to their systems. It just so reflects a consumer's expectations.</i></sarcasm>

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>
May 14, 2003 2:39:06 PM

Quote:
Everyone keeps saying that the AXP3200+ should have been 2.4-2.5GHz or should have been an AXP 2900+. I don't entirely disagree from the perspective of my desires, but I am also completely unsurprised that AMD didn't do this from a business perspective.

Frankly, the 2.4GHz AXP3200+ would have been a ludicrous thing for AMD to do for a few reasons:
1) It would leave them with no further headroom at all to try and counter Intel with later. (Because that pretty much is the limit of the current Athlon die revision and I don't see AMD bothering with the Athlon anymore until they get their .09micron process down.)
2) The yields on 2.4GHz (or higher) AXPs that could actually be sold as such (and not as special overclocked chips) would probably be so bad that they'd have hardly any per batch. Most AXPs won't get there without serious cooling and a core voltage increase. The percentage of AXPs that can hit 2.4GHz with their retail cooling and stock voltage is miniscule. We like to rag on Intel's latest heat-monger CPUs, but the AXPs at 2.4GHz and higher would be putting out just as much (or really more) heat, but they have a significantly smaller amount of surface area on their die to transfer that heat to the heat sink. The dies are just simply too small. (And rectangular at that, which kills surface area. Square would be ideal.)
3) AMD's PR rating is already set on a certain path. Now, I may not fully agree with that path, but that's not really the point. The point is that for AMD to suddenly jump the processor speed just for a 200 rating increase would really screw up what little shred of sanity is left to their rating system. Just as for them to lower the ratings on any of their CPUs at this point would severely screw that up as well. AMD laid their ratings path out and now they have to follow it.

So honestly, I don't have any quarrels with AMD over the PR/MHz that they gave the chip. I may not agree with their PR scheme in general, but the chip fell perfectly into their scheme with much less confusion than some of their others, so I can't fault them for that at all.

It's unfortunately pretty obvious or seems obvious that AMD wants the Athlon XP line to compare with the P4. AMD says that they're comparing it to the Thunderbird core, which due to a few reasons, seems pretty unlikely. For example, since the Athlon XP 3000+ is clocked at 2.16 Ghz, and the AXP 2800+ is rated at 2.08 GHz, one may wonder what those extra 200 PR points stand for.

However I understand your points perfectly, and I agree with them too. If AMD does release a 3300+ CPU clocked at 2.5 Ghz, it would make them look stupid indeed, not a single doubt. But since that many people is of the belief that AMD wants to rate the AXP towards Intels P4 instead of AMDs own T-bird, rating the 3200+ at just that, looks stupid as well. It's hard to say what AMD needs to do. Probably it's as you say, they would preferably keep their PR rating for now, at least until Athlon 64 comes out. Then, they should remove the PR rating, if you ask me. If they do if when they have a new product out, I don't think they'll lose their creditibility, which they probably would, if they removed in all of a sudden on the entire Athlon XP line of products.

Quote:
Not likely. If anything I made an observation based on your posts. There is a considerable difference between observing and believing. Observations change whenever that which is being observed changes. Beliefs change only when our own personal judgements change. There is only one person in all of THGC that has been so consistant and memorable that I hold any beliefs on, judgements of. Everyone else I just make observations of, which hold no judgement and therefore change anytime the posts change.

In the beginning when I was new here to THG Community (have been visiting THG for several years though), I made that stupid post that made me, indeed look like a stubborn and very diehard AMD fanboy. Admittedly. But then when I discussed something else with you, I think it was about overclocking and further clock speed increases of AMD and Intel, then I thought you'd rule me out as a close-minded fanboy once again, when I felt I really tried to be open-minded which I mostly feel like I am. Sure, I might have overreacted or you might have had a bad day, but anyways, I am *not* a close-minded stubborn AMD fanboy. I try to be open-minded about things and to hear others opinions. I'm not sure what "beliefs" you say you have of me, but I'm happy to listen to what you have to say about it, if you can remember it. I am, like most people, able to see my mistakes, and learn from them. So if you have anything bad you remember from me, please, don't hesitate to tell me.
I certainly is someone who don't like to have any "unfinished business" with anyone. You might have noticed. I also might be overly sensitive. I have to work on that. By the way, here is the link to the thread where you did not "finish" the discussion.

http://forumz.tomshardware.com/hardware/modules.php?nam...

Quote:
That's entirely possible. Like any imperfect situation, I cannot be here 24/7/52* and as such some conversations become lost to me. Such is real life. I do not pretend to make the pretense that I will always be here to continue every conversation. Frankly, I just don't have that kind of time and as such I have to prioritize what I have.

Of course, you can't and wouldn't want to be here 24/7. Neither would I. The only thing I wanted to say about this was that you felt ignorant of my post. But if it wasn't done on purpose, then it's no big deal at all. Of course, you do what you want, it's not that. It's just that I want to clear this up, if possible.

Quote:
When I am not self-animated, I am merely the mirror which through you may see yourself. If you are creative and open-minded, my replies will be creative and open-minded. If you are not, I am not. It is simply a matter of giving back the same respect given, which is exactly my modus operandi.

I have to ask though, which discussion are you referring to? This one or whichever one that I have not continued in the past?

Certainly. I mean nothing else than to try to give the best respect to anyone. To give is to get and to get is to give. No doubt. All points correct.

The link to the discussion is above.


My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 14, 2003 4:09:42 PM

Sometimes these posts just get too damn long to read.
May 14, 2003 4:14:48 PM

It was meant as a reply to slvr_phoenix.

My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 14, 2003 4:39:54 PM

Quote:
It's unfortunately pretty obvious or seems obvious that AMD wants the Athlon XP line to compare with the P4. AMD says that they're comparing it to the Thunderbird core, which due to a few reasons, seems pretty unlikely. For example, since the Athlon XP 3000+ is clocked at 2.16 Ghz, and the AXP 2800+ is rated at 2.08 GHz, one may wonder what those extra 200 PR points stand for.

I completely agree. AMD's take on their rating reasoning is highly unbelievable. My first and still strongest argument is to take it right back to the beginning. Compare the Palomino 1500+ to the 1400MHz T-Bird. The rating system never made sense from the very beginning. (And I might add, the 200 points for the AXP3200 don't really make sense now. A hundred would have been far more accurate.)

Quote:
It's hard to say what AMD needs to do. Probably it's as you say, they would preferably keep their PR rating for now, at least until Athlon 64 comes out. Then, they should remove the PR rating, if you ask me. If they do if when they have a new product out, I don't think they'll lose their creditibility, which they probably would, if they removed in all of a sudden on the entire Athlon XP line of products.

That is exactly how I see it. Hopefully the Athlon64 will no longer implement a naming scheme that pretends to look like MHz. (And for that matter will not even pretend to be useful for comparing the performance of one Athlon64 to another.) For AMD to change from their current course now would cause a loss of face that is considerably larger than the one that they are getting for staying on their present course. It is the lesser of all possible evils, though not particularly the most moral/honest.

Quote:
I'm not sure what "beliefs" you say you have of me, but I'm happy to listen to what you have to say about it, if you can remember it. I am, like most people, able to see my mistakes, and learn from them. So if you have anything bad you remember from me, please, don't hesitate to tell me.

I remember, and I don't hold any beliefs about you. As I said, there's only one person here that I do, and that's xxsk8er101xx, and that's only because he has been very consistant in his behavior and quality of knowledge. For everyone else I work purely on an observation by observation basis. I don't take anything personally or prejudice myself against people... with the one noted exception, and even that could change were he to.

In fact some of the people that I get along best with here are also some of the people that I have gotten along worst with here in the past and vice-versa. Such is the ebb and flow of life and the joy of individuality. It's what makes living fun. :) 

And I'll revive the other post, but I can't promise to always reply.

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>
May 14, 2003 5:18:55 PM

No, you misunderstood my intent.

If AMD sells a product rated at "3000+", it should perform as such. As a consumer, I don't care what is inside the chip that makes it so fast.

The fact that Intel requires SSE2 optimized code makes no difference. If the AMD chip is labeled to be competitive with the Intel chip, SSE2 or not, the AMD chip should run the same code in the same time. If not, then the PR rating is misleading to the consumer.

<font color=blue>This is a Forum, not a playground. Treat it with Respect.</font color=blue>
May 14, 2003 6:20:58 PM

Quote:
No, you misunderstood my intent.

You're right, jlbigguy. I did misunderstand what you were saying. I thought that you were trying to say that it was unfair that one processor supported a code extension that the other doesn't, but I see now that I was wrong. You were just saying that what matters is how well the software gets run, no matter how the CPU actually internally runs it. I apologize for misinterpreting your words.

Quote:
the AMD chip should run the same code in the same time. If not, then the PR rating is misleading to the consumer

I agree to an extent. I agree in that I think AMD's PR rating is in fact misleading customers in just this manner that you describe. However, I disagree in that it is too difficult to actually prove that AMD is doing so on purpose. It is possible (however unlikely) that it was really just an accident and AMD really did in fact mean it as a comparison to their T-Birds, not to Intel's P4s. But then again, we get back to that the reality of it is that it really is misleading customers, whether intentional or not, it <i>is</i> misleading. So I guess I do agree with you completely after all. Heh heh.

<font color=purple><pre><b>There are 10 types of people in this world: those who can understand binary and those who can't.</b></pre><p></font color=purple>
May 14, 2003 6:54:29 PM

I find it facinating that Intel has to have such a commanding lead in the MHz number but only amounts to a few percentage points in the real world. If AMD has to use the PR thing to sell cpu's. Why the hell not. If you took the 3200+ and ran it against a real 2.2 P4 it would be a blowout. So calling it a 3200 seems better than calling it a 2.2 The main reason for this is the amount of uneducated public out there that just see the numbers and don't know what they mean. If you want AMD to stop using their PR system, go out and spread the word what the numbers really mean.
May 14, 2003 7:56:07 PM

Using fair and correct PR numbers is one thing. Using false and misleading PR numbers is something completely different. The Athlon XP 3200+ CPU can't even beat the P4 3.0 so why is it rated at 3200+? To make AMD look as good as Intel, nothing else. If it's running slower than it's P4 counterpart is obviously not important. If AMD wants to use it's PR system, which is a pretty dumb idea after all, since even where an AMD processor generally corresponds to it's Intel counterpart, it's not absolute science. Actually the Athlon XP3200+ would lose out against let's say a P4 2.53 in a handfull of benchmarks.
Some benchmarks run better on AMD CPU's, and some run better on Intel CPU's. That's why a rating system is always a misleading idea. However AMD has recently taken it one step longer into the wrong direction, because with the 3200+ for example, the CPU won't even generally compare with the P4 3.2.

However you're right about one thing. The uneducated market doesn't know that Mhz doesn't always mean performance. Thus AMD created it's Performance Rating system, to compare against Intel. But it's still a mediocre way of doing it as I explained above.

Finally however Intel doesn't do right either when releasing a CPU mentioning just the MHz in it's name besides the actual name of CPU itself. But Intel doesn't state MHz = Performance, after all. Intel state speed, not performance. AMD states with eg. 3200+, both speed *and* performance. That's what AMD does with their PR rating.

I think AMD is starting to go too far with their understandable wish to keep apace with Intel. Just go to AMD's website and watch every new AXP release they make. With the 3200+, it goes like this: "AMD releases Athlon XP 3200+, the World's Highest Performing PC Processor."

This is just stupid. Not even Intel claims their CPU's are the best performing. At least not officially on their website. AMD needs to rethink their strategies. It's a damn fine company but lately they've been screwing up.


My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW<P ID="edit"><FONT SIZE=-1><EM>Edited by sabbath1 on 05/14/03 04:03 PM.</EM></FONT></P>
May 15, 2003 8:30:46 PM

Damn it who destroyed the word wrap?! I hate it when I have to scroll to the right for the full message!

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden. :smile:
May 16, 2003 12:26:39 AM

Had to be sabbath1 with that super-monstrous non-linking linky thing. :frown:
May 16, 2003 10:30:30 AM

Sorry, but a long message requires a long answer ;) 

My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 16, 2003 10:31:09 AM

Sorry, must've been me. Will do better next time ;) 

My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 16, 2003 10:05:50 PM

Why does everyone always say that AMD's PR ratings are misleading, but no one mentions that Intel does exactly the same thing with the Celeron. A 1.7 Ghz Celeron will never be the same as a 1.7 Ghz P4 or an XP1700+


__________________________________________________
It's not important to know all the answers, as long as you know how to contact someone who does.
May 16, 2003 10:33:04 PM

The problem with this is that AMD's Performance Rating is just that. A rating. Rating translates into displayed speed _and_ performance. What Intel does is only to display the speed, not performance. Celeron 1.7 means that Celeron CPU runs at 1.7 Ghz. P4 1.7 means that P4 CPU runs at 1.7 GHz. Thus it displays just the speed, not the performance. You with me on what I mean?

Because of the fact that some programs run better on AMD CPU's and some run better on Intel CPU's, it is nigh on impossible to set a any rating on any CPU. That's how, at least, I see it.


My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 17, 2003 2:56:17 PM

I understand your point, but this "truth" is equally or more misleading to the general public than AMD's PR. The average person thinks that a 1.7Ghz Celeron has the same performance as a 1.7Ghz P4 or an XP1700+.
And if this person understands the difference between the Celeron and the P4, then he will surely understand the difference between the XP and the P4.


__________________________________________________
It's not important to know all the answers, as long as you know how to contact someone who does.
May 17, 2003 3:12:35 PM

I understand your point as well. But if a person who doesn't know anything about computers and are on his/her way of purchasing a new CPU, and comes to choose between the Celeron 1.7 and a Pentium 4 at 1.7, I think will understand that two products using a different name and with two different price tags, will have some clear differences in them. However if the same user also had AMD in mind, (I know these choices are more than a bit more high-end but I think you get the picture). Anyways, if the same user had choices of XP2500+ and XP3200+, then the user would surely expect to get 700 mhz better performance out of the 3200+ and also that the products would compare against the P4 3.2. When this ain't the truth then that's very misleading to the customer.

While I agree completely on that Intel may be a bit confusing as well with a product named "Celeron 1.7 etc), I still think AMD's PR is a bit more misleading. But Intel is misleading as well, no doubt.


My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 17, 2003 5:58:19 PM

Quote:
The problem with this is that AMD's Performance Rating is just that. A rating. Rating translates into displayed speed _and_ performance. What Intel does is only to display the speed, not performance. Celeron 1.7 means that Celeron CPU runs at 1.7 Ghz. P4 1.7 means that P4 CPU runs at 1.7 GHz. Thus it displays just the speed, not the performance. You with me on what I mean?

Absolutely agree with you on this. And I also agree that even a layman probably knows that if there´s a P4 1.7Ghz and a Celeron 1.7Ghz at different prices, there´s probably something different between the two.

The problem here, of course, is AMD´s PR rating alone. It´s very hard to keep going with that... So we have to cut AMD some slack. However, the 3200 is so terribly overrated that I don´t know if I can find in me the good will not to be upset...
May 17, 2003 6:39:32 PM

There's bound to be some confusion about that different Intel CPU's run at different MHz, yet the slower CPU can perform better than the one running at a higher Mhz (eg: Itanium 2 VS. P4). But I still think that Intel is displaying only the speed. There's no real telling from Intel themselves that MHz is the one factor to performance.

Yeah hopefully Athlon 64 will be free of any ratings (though this seems unlikely). They just can't keep adding unrealistically high numbers to their rating every time they release a new CPU. I mean, the XP2800 Barton was at 2.08, and the 3000+ was at 2.16. 80 Mhz+ = 200 MHz performance? C'mon... And XP3200+ at 2.2 GHz, 400 FSB. VS. XP3000 at 2.17, 333 FSB. That don't make 200 extra performance either.

Sadly, AMD's competitiveness in the desktop sector is fading really fast.


My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 17, 2003 7:49:05 PM

boy, all this sure makes good reading..

somebody stop me!!
May 17, 2003 9:52:11 PM

Indeed... Intel never actually brainwashes in that sense.

Just one more thing: IA-64 CPUs are quite some beasts in clock-per-clock, huh? Just look at those floating point calculations... :cool: a sub-Ghz Itanium can take on a 3Ghz Pentium 4!... not bad. Banias has good IPC as well. I hope Prescott improves on the not-so-stellar P4 IPC...
May 18, 2003 12:23:43 AM

The Itanium 2 indeed seems to have a killer IPC. If it was 32-bit compatible you know, it could almost have been a choice of purchase. Well maybe not completely but somewhat close :p  Actually the upcoming IA-64's will have one 1.3 Ghz CPU with 3 mb L3 cache much like the current top of the line, the 1 Ghz, yep also with 3mb L3. But the coming new CPU will have a record low cost of around $1100, from what I've heard. While of course the Madison with 1.5 GHz at 6 mb L3 cache retains the high-price of $4000 of the current top-of-the-line IA-64.

One thing is for sure. It's a processor to dream about ;) 


My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW<P ID="edit"><FONT SIZE=-1><EM>Edited by sabbath1 on 05/17/03 08:26 PM.</EM></FONT></P>
May 18, 2003 5:59:04 AM

Deerfield will debute at 1 GHz with 1.5 MB of L3 cache and will cost around $700 as I recall the pricing scheme. Possibly less. I suppose it is placed to be in direct competition with Opteron and will replace the Xeon MP's for Intel's mid-low end server processors.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
May 18, 2003 9:55:23 AM

At $700, Itanium 2 would be affordable to lot more people. I wonder what the mainboard costs?

My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
May 19, 2003 2:37:40 PM

And if it will still have favorable x86 performance, for the transitional crowd. I think Opteron may still beat it there.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden. :smile:
May 19, 2003 4:36:15 PM

Perhaps the best way to state it is, an AXP pr rating is double misleading, it lies about it's speed and performance (at the high end). Celery just lies about it's performance.

Shadus
May 19, 2003 4:56:52 PM

Yeah Opteron could stand a chance. And it'll be exciting to see if it does.

My system: AMD Athlon XP 3000+ CPU / TwinMOS 1Gb DDR400 / Soltek 75FRN-RL /
Sapphire Radeon 9800 Pro / Antec True Power 550W / Maxtor 80Gb ATA-133 / Hercules GTXP SC /
Samsung DVD / Lite-On CDRW
!