Worst Anandtech review evar

[lookouthere]

if this is to test desktop usage, then why using workstation benchies?
this should test XEON vs. OPTERON and enable NUMA on OPTERON
isn't that right?

 

[DAPUNISHER]

Originally posted by: lookouthere
if this is to test desktop usage, then why using workstation benchies?
this should test XEON vs. OPTERON and enable NUMA on OPTERON
isn't that right?

Got any desktop 64bit software we can run? There just isn't enough windoze stuff for AMD 64bit let alone Intel's implementation to test 64bit capabilities with. The 'nix crowd might be able to point out some problems with the distro chosen, and other's the benchies, but it seems like a reasonable test setup for 64bit comparison. My only gripe is the 3500+ being chosen, it's more a whipping boy for, than true competitor against, the 3.6ghz Nocona IMHO. EDIT: Performance rating bullshiat aside

 

[lxie123]

As for comparing the two CPUs, well, neither P4F 3.6 nor Xeon 3.6 are out on the market so why bother doing such a rushed job? Is a job QUICKLY done better than a job WELL DONE?

Xeon 3.6 is available (at least anandtech has gotten their hands on one) while p4 3.6F is not. Otherwise I'm sure they would of conducted the article with a P4 64bit CPU.

Originally posted by: DAPUNISHER

He's using the Xeon Nocona 3.6 because they don't have the P4 3.6 with EMT64 right now.
In this review, the author is trying to compare 64 bit desktop performance and subsituted a nocona for p4 3.6 with emt64 due to lack of availability. I don't think his intent was to do workstation/server review and then subsituted athlon 64 3500+ for the opteron 150/ 250 like everyone is accusing him of.

The logic in this is still lost on me though Why pick the 3500+ for any reason??? It simply isn't the right CPU to match up even if attempting to look at AMD vs Intel in 64bit desktop performance. If attempting to compare what would be Intel's best 64bit desktop CPU against AMD, then use AMD's best Desktop 64bit CPU, and it most certainly is not the 3500+.

I see the logic in comparing P4 3.6F to A64 3500+. Since AMD called it 3500+ they are saying it should have equivalent or better perfromance then a P4 3.5 with 64-bit .. which is a 3.6F .. I personally would like to see those results as well as P4 3.6F against A64 3700+, 3800+, fx-53 etc.


For those wanting to see nocona 3.4 vs dual opteron 250 go here.

Link

 

[MDE]

If they used a P4 3.2 and an Opteron 250 you'd all be gawking at how fast the Opteron is.

 

[DAPUNISHER]

I see the logic in comparing P4 3.6F to A64 3500+. Since AMD called it 3500+ they are saying it should have equivalent or better perfromance then a P4 3.5 with 64-bit

No they aren't. Since when does AMD PR their chips for Intel ones that haven't even been released at the time of the PR being arrived at? No offense intended but that's rubbish, and everyone knows that AMDs' PR scheme is all over the place and subsequently almost worthless for comparing their own CPUs, not to mention to Intel's.

 

[lxie123]

They started with ATHLON XP and they've defended that their naming scheme everytime on the basis that their AMD 1800+ xp while clocked at 1.5 etc would rock the P4 1.8A in performance, hence it was ok for them to call it 1800+ even though the true clock speed was much lower.

So, I see no reason why it's outrageous to compare a P4 3.6GHZ w/ 64-bit extensions vs Athlon64 3500+

 

[MDE]

Originally posted by: DAPUNISHER

I see the logic in comparing P4 3.6F to A64 3500+. Since AMD called it 3500+ they are saying it should have equivalent or better perfromance then a P4 3.5 with 64-bit

No they aren't. Since when does AMD PR their chips for Intel ones that haven't even been released at the time of the PR being arrived at? No offense intended but that's rubbish, and everyone knows that AMDs' PR scheme is all over the place and subsequently almost worthless for comparing their own CPUs, not to mention to Intel's.

Since they have a 3500+ and 3800+ chip out when Intel doesn't.

 

[DAPUNISHER]

Originally posted by: MDE

Originally posted by: DAPUNISHER

I see the logic in comparing P4 3.6F to A64 3500+. Since AMD called it 3500+ they are saying it should have equivalent or better perfromance then a P4 3.5 with 64-bit

No they aren't. Since when does AMD PR their chips for Intel ones that haven't even been released at the time of the PR being arrived at? No offense intended but that's rubbish, and everyone knows that AMDs' PR scheme is all over the place and subsequently almost worthless for comparing their own CPUs, not to mention to Intel's.

Since they have a 3500+ and 3800+ chip out when Intel doesn't.

:roll:

 

[DAPUNISHER]

Originally posted by: lxie123
They started with ATHLON XP and they've defended that their naming scheme everytime on the basis that their AMD 1800+ xp while clocked at 1.5 etc would rock the P4 1.8A in performance, hence it was ok for them to call it 1800+ even though the true clock speed was much lower.

So, I see no reason why it's outrageous to compare a P4 3.6GHZ w/ 64-bit extensions vs Athlon64 3500+

prove it Then prove the current PR is meant to compare to a Intel 64bit CPU. Good luck.
BTW, you don't think it's outrageous to compare a $850 CPU to a $350 CPU, well I think that we are too far seperated by personal perception of what a "fair" Intel vs AMD CPU comparison is to have any constructive dialogue come from this.

 

[lxie123]

prove it Then prove the current PR is meant to compare to a Intel 64bit CPU. Good luck.

Just google Athlon XP naming scheme

http://www.theregister.co.uk/2001/10/10/amds_athlon_xp_numbering_scheme/
"AMD has a point when it claims that megahertz alone are no longer a sufficient measure of the raw speed of a processor - something PowerPC fans have been rightly pointing out for quite a while now - and the company is right to seek an alternative way of stressing the fact. Our concern is simple: we're not sure that slapping '1800+' on a CPU enough to convince punters that it's faster than a 1.8GHz Pentium 4.

Make no mistake, that's exactly what AMD is trying to say. For all its claims that the new model numbers signify the Athlon XP's performance relative to the old Thunderbird Athlons - purely for legal reasons; it doesn't want the Federal Trade Commission and Intel on its back - AMD's sample OEM advert compares the 1800+ XP to "competitive 1.8GHz PC processors". Since AMD doesn't actually make a 1.8GHz processor, T'Bird or not, competitive or otherwise, the chip company clearly has its arch-rival in mind."

BTW, you don't think it's outrageous to compare a $850 CPU to a $350 CPU, well I think that we are too far seperated by personal perception of what a "fair" Intel vs AMD CPU comparison is to have any constructive dialogue come from this.

how do you know the P4 3.6F will retail at 850$??? The Xeon 3.6 does but it's only being used in the review as a representative for the P4 3.6F. This is what i've been saying in the past 3 posts if you read clearly.

 

[flashbacck]

Originally posted by: lxie123
They started with ATHLON XP and they've defended that their naming scheme everytime on the basis that their AMD 1800+ xp while clocked at 1.5 etc would rock the P4 1.8A in performance, hence it was ok for them to call it 1800+ even though the true clock speed was much lower.

No... They said the AXP 1800+, while clocked at 1.5, would compare with the old AMD T-Bird clocked at 1.8, hence it was ok for them to call it 1800+. But, of course, they probably knew people would end up comparing it with Intel's P4.

Anyway, I agree, it's retarded to compare processors that don't cost anywhere near each other. If the P4 3.6F prices the same as an A64 3500, well then, I'll wait for that review.

 

[Xenon14]

Originally posted by: DrMrLordX
What were they thinking? An Athlon 64 3500+ vs a 3.6 ghz Nocona? Were they trying to save face for Intel after that damning Doom III article?

Where's the Opteron 150 vs 3.6 ghz Nocona review? Or hell, use an FX-53 or something, jeez.

My thoughts exactly.

 

[DAPUNISHER]

Originally posted by: lxie123
"prove it Then prove the current PR is meant to compare to a Intel 64bit CPU. Good luck."

Just google the Athlon XP naming scheme

"BTW, you don't think it's outrageous to compare a $850 CPU to a $350 CPU, well I think that we are too far seperated by personal perception of what a "fair" Intel vs AMD CPU comparison is to have any constructive dialogue come from this."

uh how do you know the P4 3.6F will retail at 850$??? The Xeon 3.6 does but it's only being used in the review at a representative for the P4 3.6F. This is what i've been saying in the past 3 posts if you read clearly.

flashbacck is right, and the reg makes an analysis and you call that proof? :roll: Unless you find AMD stating it you have no proof only conjecture, inference, and opinion.


Yeah, I know you continue to state it's merely meant to represent the F's performance. However, did KK say that? or is this just your speculation? What is the expected price of the 3.6ghz F? will it have the same cache and features? If not what will be missing or deactivated in the F? All those questions have bearing on wether or not the Nocona is even being used in the manner you suggest, and his conclusion contradicts your presumption right here

That's not to say that the Xeon CPU necessarily deserves excessive praise just yet. At time of publication, our Xeon processor retails for $850 and the Athlon 3500+ retails for about $500 less. Also, keep in mind that the AMD processor is clocked 1400MHz slower than the 3.6GHz Xeon. With only a few exceptions, the 3.6GHz Xeon outperformed our Athlon 64 3500+, whether or not the cost and thermal issues between these two processors are justifiable.

That would seem to definitivly state that he is directly comparing the two and not using it as a stand in. So do I except your speculation when evaluating this review, or the reviewer's own words? Tough call, not :roll:

Now, Here's a twist for you. Since you are using AMD's PR against them in this review, Claiming it is OK because AMD says it should perform like a 3.5ghz Intel, then turn about is fair play Intel claims HT is a god send to multitasking, creating 2 "virtual CPUs". Since I can add another 3500+&board for less than that Nocona costs, it would just be "fair" to test the 3.6ghz with HT against 2 3500+ systems ability to get work done.

Here's the senario, one 3500+ system does one of KK's tests while the other does one as well, while the 3.6ghz w/HT enabled is forced to run both those tests with only HT to help. Now, which platform do you think will win most the tests? The 3.6ghz HT or the 2 3500+'s handling a single task each. If the 3500+'s win the majority of the tests then is it fair? :roll:

 

[lxie123]

"Although the Athlon 64 3500+ and the Xeon 3.6GHz EM64T processors were not necessarily designed to compete against each other, we found that comparing the two CPUs was more appropriate than anticipated, particularly in the light of Intel's newest move to bring EM64T to the Pentium 4 line. Once we obtain a sample of the Pentium 4 3.6F, we expect our benchmarks to produce very similar results to the 3.6 Xeon tested for this review. "

That quote makes me think the xeon is a stand in.

and the author's message below explains why he didn't use a opteron or a better a64 for now.
"We will benchmark some SMP 3.6GHz Xeons against a pair of Opterons in the near future, so check back regularly for new benchmarks!"

Bottom line: I don't think it's so bad for him to give us a taste of how P4 w/ emt64 may compare to the a64 3500+ by subsituting with a XEON nocona 3.6 in single processor mode.

will it have the same cache and features?

from what i've read, it will have the same cache and features of the current P4 3.6E except in addition it will have EMT64
Specs

 

[BG]

thread in Ace'sexplaining the testers fault.

http://www.aceshardware.com/forum?read=115093869

 

[jdogg707]

The review was crap, PR rating or not, using are comparing two chips that aren't even in the same class, a server Xeon chip vs. a mainstream desktop chip! If they wanted fair comparison they should have put in an Opteron at 2.2GHZ (148?). This must have been slapped together quickly from two machines they already had going, because after reading it, I had to think to myself, did Intel coerce Dell into give them a big advertising deal with the agreement that they would play up the next P4? I am all for the P4 being faster, if it is then great, but comparing two chips that have nothing to do with each other in a review, and calling it fair is just wrong.

 

[DAPUNISHER]

I'm not trying to bust your chops lxie123 You have an opinion and you are entitled to it. Given the context of this review and the lack of qualifying remarks about it in the conclusion, I won't give KK a pass on this, and neither are most the other members here evidently. The PR comments are still just rubbish IMO, and I think the HT senario helps point out why it is hypocritical\bias to base a review off of it vs a 3.6ghz $850 CPU.

Your points about the 3.6F are questionable simply because for the immediate future it will be up against comparably priced AMD 64bit CPUs in a 32bit windoze enviroment and these benchies will have no relevance in that situation. Why? Because most desktop CPUs aren't used for most of those tasks, so the normal 32bit benchmark suites used by hardware sites will be where the 3.6F has to perform well. Unless it can be made to run MP then most of what was tested just doesn't mean squat to most potential 3.6F desktop owners.

Anywho I have enjoyed being exposed to a radically different perspective, I just can't fit in the shoes so "I can walk a mile in them before judging the owner" is my problem

 

[Dug]

thread in Ace'sexplaining the testers fault.

http://www.aceshardware.com/forum?read=115093869

Everyone should read this thread.

 

[Megatomic]

I can't wait to see KK or ANY AnandTech staff respond to this thread. I'm confused as hell about this review right now.

 

[lxie123]

Your points about the 3.6F are questionable simply because for the immediate future it will be up against comparably priced AMD 64bit CPUs in a 32bit windoze enviroment and these benchies will have no relevance in that situation. Why? Because most desktop CPUs aren't used for most of those tasks, so the normal 32bit benchmark suites used by hardware sites will be where the 3.6F has to perform well. Unless it can be made to run MP then most of what was tested just doesn't mean squat to most potential 3.6F desktop owners.

In that case people should complain about the tests being run rather than whether or not it was a fair comparison to choose the processors in question. I was reponding to the complaints about the later. The author believes that the Xeon is a good representative of how a P4 with EMT64 would perform under these tests.

 

[DAPUNISHER]

Originally posted by: lxie123

Your points about the 3.6F are questionable simply because for the immediate future it will be up against comparably priced AMD 64bit CPUs in a 32bit windoze enviroment and these benchies will have no relevance in that situation. Why? Because most desktop CPUs aren't used for most of those tasks, so the normal 32bit benchmark suites used by hardware sites will be where the 3.6F has to perform well. Unless it can be made to run MP then most of what was tested just doesn't mean squat to most potential 3.6F desktop owners.

In that case people should complain about the tests being run rather than whether or not it was a fair comparison to choose the processors in question.

Most are! Now that I read that link to Ace's I am starting to see why! It looks like this was a real SNAFU all around given the current feedback on it by others.

 

[DrMrLordX]

AMD's PR rating system is trash. Rubbish. GARBAGE. NO useful comparison of AMD CPUs can be made based solely on their PR rating. It's nothing but a marketting number! Case in point: the 3500+ runs at the exact same clockspeed as the Newcastle 3200+ for socket 754. It has the same size l2 cache, too. Hell, the core is pretty much the same, except that the 3500+ has dual-channel memory support on the socket 939 platform. Previous tests have shown that dual-channel memory "ain't all that" for Athlon 64s. The performance advantage of the 3500+ over the 3200+ is marginal, at best.

If this article was intended to show the weakness of the AMD PR system, then it is redundant. Previous articles have done the same rather nicely. There was no reason to do so in such a heavy-handed fashion.

For a "fair" benchmark, you have to take an AMD cpu that fills the same market segment or would otherwise be expected to compete against a hypothetical 3.6F P4. Namely, you want:

Opteron 150
Athlon 64 FX-53
or
Athlon 64 3800+

Those will be competeting against Intel's top-of-the-line cpus in the very near future, simply because those are competing against Intel's top-of-the-line cpus right now. The 3500+ is the current "mid-range" and/or "value" cpu for the 939 socket. In my opinion, the 3500+ is currently horribly overpriced, considering how cheap the 3200+ and 3400+ are for socket 754. In many cases, the 3400+ is a faster cpu, to boot.

On NewEgg.com, a 3200+ Newcastle is $215 retail. The 3400+, which is clocked 200 mhz higher than the 3500+, is only $294 retail. The 3500+ is $352, even though it has the same specs as the $215 3200+(aside from the irrelevant dual-channel memory support).

So yes, AMD's PR ratings for socket 939 are screwed up, not that you had to go to the trouble of comparing it to a 3.6 ghz Xeon to point that out. In conclusion, I must say that it was NOT appropriate to compare these two CPUs.

 

[DAPUNISHER]

DrMrLordX, Word. :beer:

 

[bunnyfubbles]

Yes well while AMD's PR system may indeed be inacurate and misleading, so is the difference in clock rate between Athlon CPUs and P4s. AMD is most likley heavily exagerating their PR for the socket 939 if only to promote it over the 754. Obviously the 3500+ tag is far more attractive to the uneducated buyer vs. a 3200+ or even 3400+ tag. And if they need reassuarnce they can ignorantly be pointed to the dual channel memory support.

Two parties are at fault here, 1st there's Intel for designing the P4 to scale high and look deceptively fast with high clock rates (because they knew people would buy up the GHz marketing) then there's AMD for trying to fight back with a rating system that doesn't seem to hold a consistent system. They probably should have devised something different but they didn't so now we're stuck where we are and for those of us who actively keep informed should know better and shouldn't be taken aback by such things including this article.

 

[Viditor]

All I can say is they should fix up the errors and address the inadequacies. And THAT should be done quickly

This is the most intelligent line on this thread! We have all (or at least most of us) come to appreciate and trust AT over the years. When they make a mistake (rarely) they admit it and correct it with an update.
Obviously, there were a number of mistakes on this review. I expect AT will correct them as soon as they have had a chance to review the review (so to speak...).

Kris, if you're reading this I highly suggest the Aces thread...between the idiotic pieces of vitriole are some very good points! In addition, if you could please post what the memory timings used were...
Also (pressing my luck here), testing with at least 4 Gigs of Ram would be more appropriate IMHO.