The 2166mhz Barton does not deserve its "3000+" PR

[SexyK]

Originally posted by: Warin

Originally posted by: Wingznut
Warin... I don't think anyone ever claimed that the P4 was designed to compete clock-for-clock. It was designed to be the fastest and most reliable desktop CPU on the market. Yes, Intel took a different path than AMD attempted to reach that end, but does it really matter how you get to Point A?

Actually, it does. Making a core that ramps really well (though I read the current core (~105w heat dissipation in the 3.06 part) may be near its limits) but on a mhz to mhz comparison doesnt even compete is hardly 'the fastest'. In mhz, yes, the P4 is the 'fastest' but using mhz as a metric for performance is actually pretty weak. In actual benchmarking, the fact that the 3060mhz Pentium 4 doesnt completely slaughter the 2167mhz Barton Athlon is actually pretty funny. Because it shows that Intel banks more on the clock speed than actual performance. Even in the SSE2 heavy benches, it's not a complete rout.

I'm not flaming you, BTW. The 3.06 P4 is a damn sweet part in its real world performance. But it doesnt convincingly kill a part that is -900- mhz SLOWER. And to me that speaks volumes. I am not quite ready to trade in my Athlon XP for a P4, but if I have to wait much past September for an Athlon 64 (Unless the Opteron isnt prohibitively expensive 😀) then I'll have a spiffy Intel Inside badge, I suspect.

edit:
Reliability: I've built 20-30 Athlon systems for friends. I've never cracked a core, and I havent (as yet) seen a single processor failure. That compares favourably with the Pentium systems that I've built. How is the P4 more reliable, other than in the extremely unlikely event that a heat sink completely comes loose? 🙂 I do like the integrated heat spreader a lot though, glad AMD finally saw the light and are integrating it on the Hammer line!

Oh my God, when will posts like this stop? The Athlon XP does more per clock cycle, but can't physically handle more clocks per second. The P4 does less per clock cycle, but is capable of running at much higher clock speeds. *I'm about to use fake numbers to make a point. Please don't expect them to have any correlation to a real XP or P4* Lets say the XP handles 10 instructions per clock, and the P4 7. Now lets say the XP can run at 20 cycles/sec and the P4 can run at 30 cycles/sec. No matter how you slice it, the P4 still completes 210 instructions/second and the XP only 200 instructions/second. Even though it does less per clock, the P4 is STILL A FASTER CHIP! What's the big problem with Intel's approach to processor design, to me it looks extremely intelligent and forward thinking. Please stop rehashing the old IPC debate - if you truly believe that GHz don't matter, then you'll look at the performance numbers forget the PR rating or clock speed - and see that the P4 is the faster processor in most instances, despite its low IPC. If AMD could build a chip with the ICP of the Barton and the clock speed of the P4, then you would have a point in saying "hey look, AMD's chip is simply technologically superior to intels" but until then, its pretty pointless to worry about the clock speed discrepency.

 

[paralazarguer]

I'm not giong to go hunting for the quote but someone said that the P4 had a superior architecture because it is the higest performing processor (implying that the AMD approach which was being discussed was good but didn't achieve the same level of performance.)

I believe that you can't really compare the different processor architectures based on what the fastest CPU on the market is. I think that a huge factor is each company's fab abilities.

AMD just recently started cranking out the good .13u chips. Let's image that it was the other way around. Let's say that AMD was using the P4 architecture and Intel had the athlon architecture. So, AMD would just now be rolling out the 2.2A or therabouts and would be laughed at for its poor IPC ration.

Intel, on the other hand, would have a really mature .13u like they do now and their Athlon XPs would be far ahead of where AMD's are today.

It's not so much that the P4 architecture is inheriently better. It's just that a higher clocked P4 can beat the performance of an athlon and overcome its IPC problems. This is dependant upon being able to manufacture chips with the newest strained silicon and very mature .13u however.

Make no mistake, if the shoe was on the other foot, the P4 architecure would flop if being made by AMD>

 

[LeeTJ]

Originally posted by: SexyK

Originally posted by: Warin

Originally posted by: Wingznut
Warin... I don't think anyone ever claimed that the P4 was designed to compete clock-for-clock. It was designed to be the fastest and most reliable desktop CPU on the market. Yes, Intel took a different path than AMD attempted to reach that end, but does it really matter how you get to Point A?

Actually, it does. Making a core that ramps really well (though I read the current core (~105w heat dissipation in the 3.06 part) may be near its limits) but on a mhz to mhz comparison doesnt even compete is hardly 'the fastest'. In mhz, yes, the P4 is the 'fastest' but using mhz as a metric for performance is actually pretty weak. In actual benchmarking, the fact that the 3060mhz Pentium 4 doesnt completely slaughter the 2167mhz Barton Athlon is actually pretty funny. Because it shows that Intel banks more on the clock speed than actual performance. Even in the SSE2 heavy benches, it's not a complete rout.

I'm not flaming you, BTW. The 3.06 P4 is a damn sweet part in its real world performance. But it doesnt convincingly kill a part that is -900- mhz SLOWER. And to me that speaks volumes. I am not quite ready to trade in my Athlon XP for a P4, but if I have to wait much past September for an Athlon 64 (Unless the Opteron isnt prohibitively expensive 😀) then I'll have a spiffy Intel Inside badge, I suspect.

edit:
Reliability: I've built 20-30 Athlon systems for friends. I've never cracked a core, and I havent (as yet) seen a single processor failure. That compares favourably with the Pentium systems that I've built. How is the P4 more reliable, other than in the extremely unlikely event that a heat sink completely comes loose? 🙂 I do like the integrated heat spreader a lot though, glad AMD finally saw the light and are integrating it on the Hammer line!

Oh my God, when will posts like this stop? The Athlon XP does more per clock cycle, but can't physically handle more clocks per second. The P4 does less per clock cycle, but is capable of running at much higher clock speeds. *I'm about to use fake numbers to make a point. Please don't expect them to have any correlation to a real XP or P4* Lets say the XP handles 10 instructions per clock, and the P4 7. Now lets say the XP can run at 20 cycles/sec and the P4 can run at 30 cycles/sec. No matter how you slice it, the P4 still completes 210 instructions/second and the XP only 200 instructions/second. Even though it does less per clock, the P4 is STILL A FASTER CHIP! What's the big problem with Intel's approach to processor design, to me it looks extremely intelligent and forward thinking. Please stop rehashing the old IPC debate - if you truly believe that GHz don't matter, then you'll look at the performance numbers forget the PR rating or clock speed - and see that the P4 is the faster processor in most instances, despite its low IPC. If AMD could build a chip with the ICP of the Barton and the clock speed of the P4, then you would have a point in saying "hey look, AMD's chip is simply technologically superior to intels" but until then, its pretty pointless to worry about the clock speed discrepency.

Just compare the ATI 9700 Pro vs the Geforce FX Ultra. The Geforce has CLEARLY won the performance battle, but not EVERYONE automatically prefers the higher Mhz given that there is virtually no difference in performance.

same idea same logic.

AMD's mistake was in coming up w/ the PR rating. they basically conceded that Mhz was the best way to judge CPU performance and all of us know that simply is not true.

 

[chilled]

Originally posted by: SexyK: *I'm about to use fake numbers to make a point. Please don't expect them to have any correlation to a real XP or P4*

Good luck with that!

Originally posted by: BlvdKing
I think the Barton 3000+ is a bit overestimated. I think the core should be clocked at 2250 MHz, the same as the 2800+, so that it would not get beat by the 2800+ when the extra cache doesn't make a difference. That would have been the best way of rating the 3000+ IMO.

I totally agree with this. In this case the AXP Barton 3000+ would possibly have been acknowledged to have been on equal terms with the P4 3.06HT if not better, leading to the continued success of their PR rating. This PR rating was particularly well accepted due to the conservative ratings, whereas now they seem more aggressive.

Maybe the 2.17GHz should have been labelled 2900+ and 2.50GHz the 3100+ if more appropriate. I dunno......but I would say this is a small misjudgement from the top guys at AMD, and the 3000+ should have been left with the 2.50GHz for marketing sakes.

 

[SexyK]

Originally posted by: paralazarguer
I'm not giong to go hunting for the quote but someone said that the P4 had a superior architecture because it is the higest performing processor (implying that the AMD approach which was being discussed was good but didn't achieve the same level of performance.)

I believe that you can't really compare the different processor architectures based on what the fastest CPU on the market is. I think that a huge factor is each company's fab abilities.

AMD just recently started cranking out the good .13u chips. Let's image that it was the other way around. Let's say that AMD was using the P4 architecture and Intel had the athlon architecture. So, AMD would just now be rolling out the 2.2A or therabouts and would be laughed at for its poor IPC ration.

Intel, on the other hand, would have a really mature .13u like they do now and their Athlon XPs would be far ahead of where AMD's are today.

It's not so much that the P4 architecture is inheriently better. It's just that a higher clocked P4 can beat the performance of an athlon and overcome its IPC problems. This is dependant upon being able to manufacture chips with the newest strained silicon and very mature .13u however.

Make no mistake, if the shoe was on the other foot, the P4 architecure would flop if being made by AMD>

Well, aside from the fact that Intel ha better fabbing facilities as a whole, the "IPC problems" you talk about are what allow the P4 to scale to such high clock speeds. If Intel were using a design similar to the Athlon XP, even their supierior fabbing facilities would not allow the chip to be running at 3.06GHz today. The chip was intentionally designed to allow its higher clock speed by lengthening the pipeline, there's no secret there. Anyway, noone said the P4 architechture was inherently better, I only said it was a different way of getting to the same ultimate goal - high performance - and you shouldn't slam it just because it does less per clock, since its capable of operating at higher frequencies.

 

[ndee]

I think Wingznut wants to spread some insecurity among the n00bs 😛

Buy what ever suits you best. End of discussion.

 

[Random Variable]

Is UMC manufacturing the Bartons for AMD? I vaguely remember reading that somewhere a few weeks ago.

 

[LeeTJ]

Originally posted by: ndee
I think Wingznut wants to spread some insecurity among the n00bs 😛

Buy what ever suits you best. End of discussion.

🙂

mb, that's the problem w/ being a known employee of Intel, no matter how objective and honest he is being, there is still the question in the back of the mind.

 

[dullard]

But if it isn't performing as well as the 2800+, then why does it deserve a 2925+ rating either?

Simply put, the 2800+ wins some and the 3000+ wins some others. So if there were never any new programs, I'd say they both deserved the same rating of 2800+. But as time goes on, the 3000+ should win in the majority of the newer programs. So maybe because of that AMD should put it one notch above the 2800+. But it certainly doesn't deserve two notches up to the current 3000+ rating.


Here is an interesting Inqurier article today:
"What's really puzzling to us is that on roadmaps the INQUIRER saw last week, AMD appeared to have downed the frequency of a Barton 3000+ from 2.25GHz to 2.17GHz"

If that is true, then AMD made a last minute switch to a lower frequency. The 2.25 GHz CPU would have easilly met its 3000+ rating, but the small jump back means they really needed to cut the rating to 2900+.

 

[First]

Originally posted by: LeeTJ

Originally posted by: ndee
I think Wingznut wants to spread some insecurity among the n00bs 😛

Buy what ever suits you best. End of discussion.

🙂

mb, that's the problem w/ being a known employee of Intel, no matter how objective and honest he is being, there is still the question in the back of the mind.

Except Wingznut makes some excellent points, doesn't matter where he's employed. 🙂

Btw, I guess Anand did use PC1066. He must have corrected his error a few hours after the article was out.

 

[paralazarguer]

Well, aside from the fact that Intel ha better fabbing facilities as a whole, the "IPC problems" you talk about are what allow the P4 to scale to such high clock speeds. If Intel were using a design similar to the Athlon XP, even their supierior fabbing facilities would not allow the chip to be running at 3.06GHz today. The chip was intentionally designed to allow its higher clock speed by lengthening the pipeline, there's no secret there. Anyway, noone said the P4 architechture was inherently better, I only said it was a different way of getting to the same ultimate goal - high performance - and you shouldn't slam it just because it does less per clock, since its capable of operating at higher frequencies.

Yean, Please don't misunderstand me. I'm not saying that the athlon would be at 3ghz. I'm just saying that because the P4 is the fastest CPU going right now, doesn't make the architecture better. Yes, the p4 has a lower IPC but it's design allows it to hit higher clock speeds (and other advantages that its architecture has) which cancel the lower IPC out.
All I'm saying is that Intel's great fabs are as much responsible for the P4 being the fastest proc. out right now as its design. If AMD made the P4, it wouldn't be at 3.06 with strained silicon even if intel designed it.

 

[SexyK]

Originally posted by: paralazarguer

Well, aside from the fact that Intel ha better fabbing facilities as a whole, the "IPC problems" you talk about are what allow the P4 to scale to such high clock speeds. If Intel were using a design similar to the Athlon XP, even their supierior fabbing facilities would not allow the chip to be running at 3.06GHz today. The chip was intentionally designed to allow its higher clock speed by lengthening the pipeline, there's no secret there. Anyway, noone said the P4 architechture was inherently better, I only said it was a different way of getting to the same ultimate goal - high performance - and you shouldn't slam it just because it does less per clock, since its capable of operating at higher frequencies.

Yean, Please don't misunderstand me. I'm not saying that the athlon would be at 3ghz. I'm just saying that because the P4 is the fastest CPU going right now, doesn't make the architecture better. Yes, the p4 has a lower IPC but it's design allows it to hit higher clock speeds (and other advantages that its architecture has) which cancel the lower IPC out.
All I'm saying is that Intel's great fabs are as much responsible for the P4 being the fastest proc. out right now as its design. If AMD made the P4, it wouldn't be at 3.06 with strained silicon even if intel designed it.

No arguement there😉 Guess that's one of the benefits of having a ton more capital to spend on your fabs.

 

[ndee]

Originally posted by: Evan Lieb

Originally posted by: LeeTJ

Originally posted by: ndee
I think Wingznut wants to spread some insecurity among the n00bs 😛

Buy what ever suits you best. End of discussion.

🙂

mb, that's the problem w/ being a known employee of Intel, no matter how objective and honest he is being, there is still the question in the back of the mind.

Except Wingznut makes some excellent points, doesn't matter where he's employed. 🙂

Btw, I guess Anand did use PC1066. He must have corrected his error a few hours after the article was out.

Of course it's true what Wingznut states, just some teasing 😛 Actually, if he never stated he worked for Intel, there would be more pro-Intel voices here I think cuz he's points are so valid hehe

 

[jbond04]

Originally posted by: Wingznut
jbond... I agree about the 1.5ghz (and slower) P4's. But nowhere did Intel give those cpu's a higher rating than their predecessor. Yes, you can say that the actual speed (in ghz) is misleading... But it's not. 1.5ghz is actually what that cpu was running. Like I said before, there have always been different cpu's running at the same speed, that didn't have the same performance.

And my statement is DEFINTELY NOT "like the pot calling the kettle black", since I don't speak for Intel. As I've already reiterated (and as if my sig wasn't enough), Intel did not post this thread. I did.

I wasn't trying to imply at all that Intel did anything wrong with the P4...I think it's architecture is greatly superior to the Athlon's, like you said, it doesn't matter how you get there, as long as you provide the performance. I really didn't mean to imply that the P4's architecture offers "misleading" performance numbers. At any rate, looking back on my previous post, it sounds a lot more accusatory than I really meant. Sorry about that. 😱 And I know you don't speak for Intel.....I was just pointing out that they have encountered a similar problem as AMD (a newer processor being outperformed by an older model at times)...I never said they "solved" the problem the same way AMD did (with the PR system).

So yeah...sorry about any animosity you might have gotten from my previous post. It was purely unintentional. 🙂

 

[LeeTJ]

Originally posted by: Evan Lieb

Originally posted by: LeeTJ

Originally posted by: ndee
I think Wingznut wants to spread some insecurity among the n00bs 😛

Buy what ever suits you best. End of discussion.

🙂

mb, that's the problem w/ being a known employee of Intel, no matter how objective and honest he is being, there is still the question in the back of the mind.

Except Wingznut makes some excellent points, doesn't matter where he's employed. 🙂

Btw, I guess Anand did use PC1066. He must have corrected his error a few hours after the article was out.

It wasn't meant to be a knock on Wingznut. but how can it not be a consideration. No one is truly objective. I don't believe in objectivity anyway, so i assume everyone has an agenda, it's easier to deal with someone who has a known agenda than someone who has a hidden agenda, but everyone has one.

 

[Wingznut]

ndee, LeeTJ, and Evan... I could've chosen to remain anonymous about my employer (as a number of other people who post here do), but I decided to be up front with people. It's the way I've been my whole life, and I'm not going to hide behind a keyboard on a msg board. 😉

And because I disclose that fact, it makes posting my opinions more difficult. I don't have the room for error that others have, ESPECIALLY when posting about AMD. If I don't have my facts straight, and present them very clearly, I would be the most flamed person here by far. 😉

That being said, I just hope that somewhere amongst my 8100+ posts I've gained enough respect and the understanding that my career is in making cpu's.... NOT selling them. 🙂


paralazarguer and SexyK... Sure, it helps to have the capital to spend on fabs... But give the talent pool some credit, too. 🙂


Vespasian... All Bartons are being manufactured in Fab30 in Dresden, Germany. I'm not certain where AMD and UMC stand now, after the agreement with IBM/AMD (their 3rd such colaboration in 4 years). I believe the partnership is dissolved, but I'm not positive.


jbond... Absolutely no animosity or implications taken... I just thought it would be a good idea to clarify that I am my own man. 😉


LeeTJ... You are correct. Nobody is truly objective. But I do think that I am less biased than a good number of people on this board. I'll tell you right now that Barton is an excellent piece of silicon. It's just the PR that I don't care for.

Hey, I was a PC enthusiast long before an Intel employee. 🙂

 

[JSSheridan]

In the future, I will be refering to the 2166mhz Barton as the '3000-'. Peace.

 

[Snoop]

I wasn't trying to imply at all that Intel did anything wrong with the P4...I think it's architecture is greatly superior to the Athlon's, like you said, it doesn't matter how you get there, as long as you provide the performance.

This is total shiet. How is it that Intels architecture is so much superior when its riveled in performance by a architecture, which I would guess has received at most around 1/15 the research and development costs?

 

[alexruiz]

Originally posted by: chilled

Originally posted by: BlvdKing
I think the Barton 3000+ is a bit overestimated. I think the core should be clocked at 2250 MHz, the same as the 2800+, so that it would not get beat by the 2800+ when the extra cache doesn't make a difference. That would have been the best way of rating the 3000+ IMO.

I totally agree with this. In this case the AXP Barton 3000+ would possibly have been acknowledged to have been on equal terms with the P4 3.06HT if not better, leading to the continued success of their PR rating. This PR rating was particularly well accepted due to the conservative ratings, whereas now they seem more aggressive.

Maybe the 2.17GHz should have been labelled 2900+ and 2.50GHz the 3100+ if more appropriate. I dunno......but I would say this is a small misjudgement from the top guys at AMD, and the 3000+ should have been left with the 2.50GHz for marketing sakes.

Another vote for that one. I think the frequency is 2.25 GHz for the 2800+, but I certainly wouldn't mind a 2.5 GHz barton..... 😉

 

[ndee]

Originally posted by: Wingznut
ndee, LeeTJ, and Evan... I could've chosen to remain anonymous about my employer (as a number of other people who post here do), but I decided to be up front with people. It's the way I've been my whole life, and I'm not going to hide behind a keyboard on a msg board. 😉

I know, no worries 🙂 and I appreciate that Anandtech has such a knowledgable person like you 🙂

 

[RaynorWolfcastle]

I don't believe this is technically correct. Just this morning, having read through a few sites' reviews of the Barton, I found, again, a statement by someone that the PR system is based on an old AMD platform. AFAIK, this is still true.

Iimater - There's what AMD claims is true and what actually happens. AMD is not comparing to the Thunderbird core as that core could have never made it to such high frequencies. I'm not sure if AMD still claims that its PR numbers are in comparison to the TBird core or not but if they do it's a totally bogus statement since it cannot be proved or disproved by anyone (incl. themselves).

This is dependant upon being able to manufacture chips with the newest strained silicon and very mature .13u however.

paralazarguer - Intel doesn't use strained silicon with it's .13 micron process, it's slated for introduction with Prescott and the .09 micron process.

AMD just recently started cranking out the good .13u chips

Actually, .13 micron or not, the Athlon is getting close to its architectural limits. Consider that the TBred A hit a wall just above 2GHz, and that AMD was (and is) having difficulty producing their 2.25 GHz part in quantities. Understand that AMD, by adding more cache, ups performance at a given clock speed the expense of making it more difficult for itself to increase frequency.

Also, someone said that AMD was waiting for MS to launch the Athlon 64 with the 64-bit Windows. Although it is possible it is very unlikely that this is the case. Heck AMD claims that expect a ~25% boost at the same clockspeed as the Tbred core. You seem to assume that AMD hit their rumoured 2 GHz target frequency. That would put them at a PR of approx 3150+ (just a guess based on the 2800+) on existing apps. The delay is more likely caused by either AMD not hitting their 2 GHz target frequency or by their partners not having motherboards ready to ship; MS having a 64-bit OS to launch with the Athlon 64 would be icing on the cake.

As to the Barton 3000+ well I think it's been pretty well explained above, AMD is stretching that PR rating very far. Hopefully they find a way to increase their core frequency to put the pressure on Intel again.

 

[Bluga]

Sure, intel's 3 GHz deserves it

We all know almighty intel rules.

 

[RaynorWolfcastle]

Originally posted by: Bluga
Sure, intel's 3 GHz deserves it

We all know almighty intel rules.

Intel markets processor with a core frequency of 3.06 GHz. What do they sell you? a processor that functions at 3.06 GHz, where's the problem?
MHz != performance. Intel does not claim performance when they sell you a 3.06 GHz CPU, AMD does because their processor does actually have a core frequency of 3000 MHz.

 

[MadRat]

I expected this sort of post from a young ATOT n00b, not from an adult. *shakes head* I'm disappointed in this topic, Wingz.

Originally posted by: Wingznut
If you choose to disagree (which I'm sure many of you will), please keep your debates rational, and your flames down to a minimum.

It would be nice to see this thread revolve around the 3000+ rating for the new Barton CPU. Unfortunately, the tone was already set with the opening post.

I'll save any comments for a more reasonable setting.

 

[Wingznut]

Don't patronize me, MadRat. I'd rather be flamed than talked to like I was an immature child.

As for the "tone" already being set, it doesn't seem as though you've read the posts. It's been very rational and flame-free.

That being said... If you'd like to disagree with me, feel free to. I welcome any rational discussions. 🙂