Xeon 3.6 vs A64 3500+, In Linux 64bit

[Kanavit]

The Xeon 3.6ghz 800mhz fsb vs AMD Athlon 3500+. Surprising results here! <A HREF="http://www.anandtech.com/linux/showdoc.aspx?i=2158&p=1" target="_new">http://www.anandtech.com/linux/showdoc.aspx?i=2158&p=1</A>

oh my, wait til the Pentium 4 3.60F gets here with EM64T. it's gonna be awesome. I can't wait myself!

<b>Conclusion:</b> "Without a doubt, the 3.6GHz Xeon trounces over the Athlon 64 in math-intensive benchmarks. Intel came ahead in every severe benchmark that we could throw at it, particularly during John the Ripper. Even though John uses several different optimizations to generate hashes, in every case, the Athlon chip found itself at least 40% behind. Much of this is likely attributed to the additional math tweaking in the Prescott family core".

------
Prescott 3.2E 1MB L2 HT
1GB PC 3200 Dual channel(PAT)
Asus P4P800 Bios 1016
PNY Geforce 6800 GT 256MB DDR3
60,823 Aquamarks

 

[ChipDeath]

Considering they'll sell at the same price point, that's a <i>great</i> comparison....

---
Epox 8RDA+ V1.1 w/ Custom NB HS
XP1700+ @200x10 (~2Ghz), 1.4 Vcore
2x256Mb Corsair PC3200LL/1x512Mb Corsair XMS PC4000 2.5-3-3-7
Sapphire 9800Pro @412/740

 

[Kelledin]

<A HREF="http://forums.anandtech.com/messageview.cfm?catid=28&threadid=1370368&enterthread=y" target="_new">Forum members review the review...</A>

...and they pretty much universally agree, the review got flubbed badly. Testing a 3500+ desktop chip against Intel's cream-of-the-crop best Xeon chip was only the start. :tongue:

Crap, at least use the FX-53 or the Opteron x50. AnandTech really didn't have an excuse <i>not</i> to, so what's up with that?

<i>"Intel's ICH6R SouthBridge, now featuring RAID -1"

"RAID-minus-one?"

"Yeah. You have two hard drives, neither of which can actually boot."</i>

 

[etp777]

I'd have to agree, the 3500+ should be tested against nothing higher than the P4 3.6 with EM64T(or a 3.4 if there is one, can't remember off top of my head).

How much that would change things, I'm not really sure(due to fact that xeon and p4 cores really aren't hugely different, or at least haven't been in past, haven't researched the EM64T chips). Whether different cache, etc would be enough to make this up, won't know until there's a better benchmark by someone.

I will be itnerested in seeing the SMP Xeon vs dual opteron review they promise.

 

[Kanavit]

there wasn't any P4 3.6F available yet, so Anandtech used a Xeon 3.6ghz with EM64T. Both are the same chips, just targeted at different price point and market segment. I thought the comparison gave us a good preview on what the P4 3.6F EM64T would give us.

------
Prescott 3.2E 1MB L2 HT
1GB PC 3200 Dual channel(PAT)
Asus P4P800 Bios 1016
PNY Geforce 6800 GT 256MB DDR3
60,823 Aquamarks

 

[Nights_L]

Why didn't they pick Opteron?

 

[Guest]

That would have been the best comparison.

Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, Leadtek FX5900 w/ FX5950U bios@500/1000, 2X30gig Raid0

 

[Mephistopheles]

You do have a point. In any case, if anything, a P4F might even perform slightly <i>better</i> than that Xeon due to DDR2-533 instead of DDR2-400. If 1066Mhz FSB came into play, it would look better. Intel is still a few steps behind anyway, though. If P4Fs would be widely available, the playing grounds would be better, but hell, they aren't.

However, EM64T is nothing laughable, it seems. I think intel should consider a high dose of alacrity when introducing the Prescotts with 2MB L2 and 1066Mhz FSB and EM64T; these will probably be great. Won't do any good if they only become available in 2005, though; AMD isn't standing still either.

But agreed; in any case, we should wait until more information becomes available.

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i><P ID="edit"><FONT SIZE=-1><EM>Edited by Mephistopheles on 08/09/04 02:13 PM.</EM></FONT></P>

 

[ChipDeath]

an Opteron would still get beaten by the Xeon, but as the point of the article is to give a 'sneak preview' of the 3.6P4F w/ EM64T, which is going to be Intel's new big hitter, they really should be comparing it with a FX-53, or 3800+, as at least these parts are actually available to the public ATM.

By the time these P4s are available in any quantity, no doubt AMD will have better stuff out too.

---
Epox 8RDA+ V1.1 w/ Custom NB HS
XP1700+ @200x10 (~2Ghz), 1.4 Vcore
2x256Mb Corsair PC3200LL/1x512Mb Corsair XMS PC4000 2.5-3-3-7
Sapphire 9800Pro @412/740

 

[darko21]

"Why didn't they pick Opteron?"

Good question. Not sure what was the point of the review. $850 chip vs $350 chip. so 250% more expensive 100% more cache, Top mhz for intel chip but not amd chip. not a good comparison. I'm thinking somone had a hissy fit over the doom3 tests. The doom3 showed all cpu's this test compared intels cream to a upper mid level amd cpu.

Reverse the situation and take an opteron or fx53 and compare to a prescott 3.4e. That would be a stupid review. I'm somewhat suprized anandtech would do that.

If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.

 

[Mephistopheles]

Not sure what was the point of the review.

How come people are having such a hard time in understanding that?... :frown:

...the point of this review is to preview P4F performance in 64-bit tasks, and it's looking good. Noone would consider either buying a 3.6Ghz Xeon ($850) or a 3500+ ($350).

This article's conclusion is that Intel's 64-bit implementation might not be flawed at all. That's it.

And P4Fs <b>don't carry a price premium</b>, so once they become more available to enthusiasts, they're a reasonable option against A64s. Even more so with all the new features (azalia, new chipsets, or even future features). Intel better move fast, though, because they seem to be stalled lately.....

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>

 

[darko21]

Re: "How come people are having such a hard time in understanding that?...
...the point of this review is to preview P4F performance in 64-bit tasks, and it's looking good. Noone would consider either buying a 3.6Ghz Xeon ($850) or a 3500+ ($350).
This article's conclusion is that Intel's 64-bit implementation might not be flawed at all. That's it."


I have no problem with showing how good p4f is or will be in 64bit tasks. HOWEVER, this statment at the end.

"Without a doubt, the 3.6GHz Xeon trounces over the Athlon 64 in math-intensive benchmarks. Intel came ahead in every severe benchmark that we could throw at it"

They are comparing intels best upcomming cpu to a mid level amd cpu. So the comment was not nessasary. If you want to make conclusions comments like that compare = cpu's.


Mephistopeles where would you draw the line on this? would it be ok in your book if anandtech did this 64bit test with a64 2800 vs 3.6 xeon?

To prove intels 64bitness is good but to compare it to a lesser amd chip like the a64 3500 is prolly not a good thing but aceptable, the conclusion however was over the top.

"Without a doubt, the 3.6GHz Xeon trounces over the Athlon 64 in math-intensive benchmarks. Intel came ahead in every severe benchmark that we could throw at it"


If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.

 

[P4Man]

>Crap, at least use the FX-53 or the Opteron x50

I have no problem with their cpu selection, remember Nocona is just Prescott, same FSB, same L2 cache, so when 3.6 Prescott with EM64T launches, that will be the exact same cpu. Comparing it to a 3500+ doesn't sound like such a terrible idea to me.

What really is f*cked up however, is:
1) they apparently messed up the scores for MySQL. They used 32 bit scores for A64 instead of 64 bit ones (like for Nocona). in reality, A64 whips the Nocona in this test
2) they used ridiculous synthetic benchmarks like SuperPI, a 5 year old mini app that tells you <i>nothing</i> or TSCP which fits completely in Nocona's L2 cache and according to Vincent Diepenveen on Aces' forum, wasn't properly compiled. If you honestly believe Xeon is better for chess programs, i also suggest you have a look <A HREF="http://www.cs.biu.ac.il/games/" target="_new">here</A> First 6 players on the computer chess worldchampionchips where opteron based. Funny for an Intel sponsored event
3) worst of all, we still don't know how intels implementation of EM64T is. Why on earth didn't they provide both 32 and 64 bit scores for both cpu's, *that* would have told us something. Now we still know nothing, unless you want to believe these results would somehow mean Nocona beats A64 hand down.

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

 

[Mephistopheles]

I agree, P4Man.

Anandtech is in fact trying to preview how a 3.6F would stack up against a 3500+. This sounds like a good idea, because both of them are comparably-rated, that's all.

However, I can also agree that the benchmarking suite wasn't chosen with care. And it is still hard to know if Em64T makes a real difference in performance...

P4Man is right, they should have done something like % increase from 32->64bits on each platform. That would have made more sense.

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>

 

[trooper11]

i dont know about you, but usually when any review site makes a comparison, preview or otherwise, they take the top models from both camps. i dont mind the review, but i dont see the logic in comparing the top model xeon versus the 'entry' level s939 part. i dont think you cna just say they are compareably rated so that means its fair, since when have reviews been done like that? top model versus top model is alot more fair, just becuase intel doesnt have a 3.8ghz xeon you think its fair to drop down to an athlon 64 3500+? i dont think so.

what really suprises me is that this is from anandtech, they are usually alot more cautious and complete wiht thier benchmarks and they always try to make a fair comparison. even a as a preview, its not a balanced review at all. not to mention the fact they did not show 32bit versus 64bit scores is another fault. it just seems kind of sloppy work for them.

the only thing i can pull form this is that it appears em64t will not be a ahndicap for intel, but it doesnt tell me wether its better then amd64, i dont see how anyone can pull that from this review. i mean goodness, they use opterons for thier webserver, youd think they might try one of them against the xeon. but oh wlel, maybe they will update the review soon.

 

[P4Man]

>top model versus top model is alot more fair, just becuase
>intel doesnt have a 3.8ghz xeon you think its fair to drop
>down to an athlon 64 3500+? i dont think so

That will depend on how the 3.6 P4 will be priced. If its comparable in price, pitching it against a 3500+ seems entirely reasonable (especially since they already had the 3500+ scores, and I can easily forgive them for not rerunning all those benches on an otherwise identical cpu with a slightly higher clock).

>the only thing i can pull form this is that it appears >em64t will not be a ahndicap for intel,

Are you sure ? I don't "pull" anything from that review, for all I know, EM64T might slow down performance on P4. Without 32 bit benches to compare, we know nothing. Unless perhaps that the P4 has a faster L1 cache therefore shines in SuperPI.

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

 

[trooper11]

again, ill ask again, shouldnt the review pit top model versus top model? they compared the xeon to an ahtlon 64, the prices between those two are even larger, i dont seem to recall review sites comparing based on price alone. sure they can say in thier remarks that people need to keep in mind a price difference, but that never excludes comparing top form top. the 3.6 xeon is thier top model, its only fair to show it up against the top amd has to offer.

 

[P4Man]

>again, ill ask again, shouldnt the review pit top model
>versus top model?

Perhaps, but consider the 3500+ does carry a rating which would imply its comparable to a 3.6 GHz p4, wouldn't you agree ?

>they compared the xeon to an ahtlon 64,

So its called a Xeon now, but who cares; it will be the exact same chip once intel sells it as a Pentium4.

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

 

[Mephistopheles]

So its called a Xeon now, but who cares; it will be the exact same chip once intel sells it as a Pentium4.

Exactly. A Nocona 3.6Ghz with EM64T is exactly equal to a Prescott 3.6F!!! That's the point, trooper11.

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>

 

[trooper11]

im not arguing that. im arguing that an opteron would have been a more fair comparison wiht the xeon, as much in price and specs.

also, a p4f 3.6 will cost roughly the same as the 3800+, if you look at intel's official pricing as of now, so why is it that the 3500+ is the better choice here? so you think its a fair comparison even if the prices arent similar and even though its intel's top versus amd's entry level for s939?

my point is this, how does anyone come to a conclusion as to which company has the best performing cpu, 64bit or otherwise? if your nto going to base it on comparable pricing and soley ont he fact that amd's pr number is similar to the 3.6ghz, then i guess your right, it is fair. but i dont agree with that. there is a 3800+ out, you can buy it right at this moemnt in quantities. its price is almost identicle to the p4 3.6, thats the non 64bit version, but we know the prices will be similar, so why leave it out?

wouldnt a 3800+ versus a p4f 3.6 show which company had the best performance? wouldnt that kind of slightly reduce any contreversy and make things more clear? im just suprsied you both feel its a fair comparison. i dont care if its xeon or p4, they are both top models, its only fair to compare them with similarly priced and top performing amd parts. im not attacking the results here, im just saying you cant make any claims based on it. not only are there too few benchmarks, but the chip selection is questionable.

i really dont think there are any alterior motives here. its pretty obvious they just wanted to get a review out as soon as possible. they already had the 3500+ scores so they owuldnt have to rerun them, so they used those and threw them out there. they just wanted to put up somehting, it wasnt one of thier complete thought out reviews. the author himself said he chose the 3500+ becuase they already had scores for it.

 

[P4Man]

>y point is this, how does anyone come to a conclusion as to
>which company has the best performing cpu, 64bit or
>otherwise? i

The article never claimed to show "which company has the best performing cpu". It was made plain clear they used a 3500+ and not a 3800+ of FX, and frankly, using a 3800+ would not have changed anything substantially. So go ahead and add 10% to the A64 scores, and what exactly changes ? Not much. And the results will still be just as flawed and useless which ought to be your main gripe with the article really.

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

 

[Kelledin]

<A HREF="http://www.aceshardware.com/forum?read=115093783" target="_new">Ace's readers speak out...</A>

...and they agree, the test was beyond flubbed. Compile options were horribly screwed up. Many figures are off by as much as 2x.

More questionable test results have been seen, but not by reliable witnesses.

<i>"Intel's ICH6R SouthBridge, now featuring RAID -1"

"RAID-minus-one?"

"Yeah. You have two hard drives, neither of which can actually boot."</i>

 

[trooper11]

oh i agree with you there, the whole thing was sloppy.

its jsut so frustrating to see ppl say things liek we should be thanking anandtech for doing these priliminary nubmers for us, its sucha good job and we should praise theme. personally, i dont praise rushed work, which is what this appeares to be. that was thier big mistake. quality over quantity, thats what i say. but of course you have ppl trying ot then blame all the rucus on 'amd fanboys'. now who here thinks thats fair?

i think the hardest thing for me is, im trying to figure out the point. i know thier intention was to show a comparison, or maybe that wasnt the intention. maybe it was just to show some random numbers.... ok see what im getting at? how can annoy figure out what they were trying to do here? they werent looking for hte best performer right? they werent looking to compare amd to intel? i just odnt see the point to this article.

if they had just waited for the full review, i owuld have no qualms, but this is just asking for trouble.

 

[Mephistopheles]

sloppy? It was even more than sloppy, it was inaccurate and uninformative. Reality is still something about which we still don't know a lot about.........

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>

 

[trooper11]

could it be that this preview was posted in order to stir up interst in the full review when it comes?

it isnt out of the realm of possibilities is it?