Post your comments and questions on the article: Intel Pentium 4 Northwood Vs. AMD Athlon XP 2000+: Pure Numbers

__________________
ActiveTuning - Partner & Director Of Sales & Marketing

No offense, but how about using all benchmarks that are not synthetic OR biased towards P4's. Everyone knows that Sandras memory benchmark will give you outragousely higher numbers with that RamBus crud, then any DDR Ram, yet real world performance shows no advantage. USe Winstone or something, at least Quake 3. Sandra is not an example of real world application. And how did the 2.3 P4 EVER score lower then the 2.2??? That doesnt make any sense.

Josh

p4-biased i might say.. i would use fav software such as quake3, 3dmark, lame mp3 encoder etc..

I agree!! And I supplied the numbers!


Sandra's memory scores are biased in favor of RAMBUS, BUT the other Sandra scores and Sciencemark's scores are not biased and the Athlon favors rather well (especially considering the Athlon is only running at 1667 MHz).

I'm sure the reason why the Prime95 scores are so bad on the Athlon, is because of the RAMBUS, though.

Fact is, if we DIDN'T use all of the same scores, someone would jump on us saying that we intentionally left out scores that favored the Pentium 4.

You can't please everyone.

I do agree that a "gaming" score should be included. Whether it's 3DMark or Evolva or Q3A... but the person that did the P4 benchmarks (someone who doesn't write for SL) was under the gun.

I asked for the scores all week and he never supplied them. He was just too busy. On the Sunday before the original numbers were posted here, I had asked him one more time to do it and thank God he did. I supplied those numbers to Dave the following Monday. I was happy to get the scores I had received that night. I wasn't going to push it. In the future, all of my benchmarks will include 3DMark 2000, 2001 and Evolva.

__________________
Hey. What kind of party is this? There's no booze and only one hooker. -- Bender (Futurama)

You know what Josh... We were trying to figure that one out ourselves.... but no one can explain why in ScienceMark the CPU ran slower than the 2.2. All other scores, the 2.3 came out on top.

My guess is because it was overclocked, perhaps the CPU was overheating and being throttled by the motherboard. That's only a guess, though.

__________________
Hey. What kind of party is this? There's no booze and only one hooker. -- Bender (Futurama)

Actually the Prime95 scores are higher due to the P4's extended instruction set that deals with precision numbers. Its stated somewhere in the Prime95 website or included readme file. The P4's actually dont give as great a precision as the XP's and so can do the calculations faster. So i guess you could say thats also a P4 biased test. Anyway, I cant wait for the .13 AMD's to come out they are gonna wipe the floor with the P4 NWood. Though I have seen a NWood O/C to ~2.65 GHz with normal air cooling, which even my AMD loving self has to admit is quite impressive. Cant wait to see what AMD has in store though, they are being awful quiet about the Th-Bred.

Josh

Well, overclocking the P4 seems to be a very simple task. That 2.0 was running at 2.3 with a regular ol' Intel fan. It actually ran faster on an Abit board, but we opted to just use the Asus we were using because it was just too much of a hassle to move everything over to the Abit after the 2.0 and 2.2 benchmarks were already done.

__________________
Hey. What kind of party is this? There's no booze and only one hooker. -- Bender (Futurama)

P4 bias I must say, sorry better luck next time.

Not just jumping to the defense of my fellow staff members, but in my mind, the pro-AMD movement is getting a little out of hand. The pendulum has swung all the way to one side.

I always read the comments on Intel vs. AMD comparisons. There was one instance in which more than a dozen people claimed a bias for Intel simply because a Pentium 4 came out ahead of an Athlon in two of sixteen tests! We should always be supporting both companies for releasing such fast CPUs. It may be much more expensive, but whatever people say, the Northwood is a FAST processor.

AMDZone has become immensely popular in recent years, while IntelZone is constantly flamed with trolls who just pop in to say that Intel can suck AMD's dick.

Intel no longer has a monopoly in the processor market, so they deserve no criticism for that. Has Intel ever forced anything harmful upon you? They may have a strong say in standards, but we gave them that power in the first place by purchasing their past processors. We aren't talking about a dictator who took the market by force, stuffing Pentiums down people's throats. Or for that matter, has AMD ever been perfect?

I own an AMD processor and love it to death. It's blazing speed and the small cost at which I obtained it amaze me. But, Intel has a great product too! Sure, they might have lengthened the pipeline of the P4 to reach higher clock speeds and win over some consumers, but they're a corporation, and that's what corporations sometimes have to do to make money. That's what corporations are here for. They make money.

When the Athlon appeared on the scene, a decade or more long history of brilliant Intel chips went out the windows in favor of the underdog. Remember that the best situation for the consumer is one in which many corporations are supported by the public.

If Intel supporters are just the quiet type, this is a time to rise up and support the other choice! I want to see raging battles on forums, not a massive AMD groopie gathering.

Talk about stepping out on a limb.

__________________
Aaron Dahlen - Gaming Editor
aaron@slcentral.com
SLCentral - Your logical choice for computing and technology
http://www.slcentral.com

Athlon XP 1800+ ... K7V DRAGON Plus ... 256 MB PC2100 ... GeForce3 PowerPack !!! ... Sound Blaster Audigy ... IBM 60GXP 40 GB ... Samsung SyncMaster 950p ... Windows XP Home

I have no problem giving Intel its dues (when they are actually DUE). People buy faster CPU's so they can...what? Run applications/Windows/IE..etc faster, right? So when rating CPU's why do we use these useless marks that mean nothing in real world performance? We have all seen the insane Sandra benchmarks that will show a PC800 P4 1.6 crush an XP2000+, but does that P4 ACTUALLY use that bandwith in ANY application? Nope. SO why even use that as a standard? Now if I am playing Quake and my new CPU gives me 20 FPS better then another, I want to know that. THAT is a resonable benchmark. If this new CPU can make my WinApps run 15% faster, I want to know THAT. If this new processor give 800000000000000 MFlops over my last processor that only gave 10 MFlops, but they both perform the same in everyday situations, does the MFlops mark mean anything??? Nope. Sure you can say that the Prime95 IS a good benchmark cause of the actual number crunching aspect, but like i stated before, the P4 actually cuts precision so if you are using it for huge PI calculations then you would actualy NOT want it, even if it was gonna be a little faster. I'm not trying to flame anyone, cause I'm glad someone is reviewing the NWood already, but I'm just saying pick and choose the benchmarks more carefully, so when I read the first review of a new CPU I can actually judge wether or not Intel really does deserve any praise for it.

Josh

Pick and choose the benchmarks more carefully so AMD always comes on top? That's what it sounds like you're saying.

I think the AMD 1667 MHz did AMAZING against the P4 2000 MHz.

The AMD actually beat the P4 2.0 GHz in Sandra's CPU benchmarks and all of the P4s in the Dhrystone benchmark and in ScienceMark. The ONLY thing that the P4s beat the Athlon at is is the RAM benchmarks in Sandra and Prime95!

If you average the numbvers, one could actually say that the Athlon really is comparable to a P4 2.0 GHz and deserves the name "2000+". So AMD did a good job with the PR naming. From the way you make it sound, the Athlon should EXCEED a 2.0 GHz comparison!! Now THAT'S biased. THAT'S lopsided.

I used to spend all my days at Anandtech's forums fighting off the Intel zealots that would call the Athlon slow, too hot, unstable. I learned then it's a losing battle. Now I'm over here fighting the same battle, but against people that thing that the benchmarks should be hand picked to favor AMD.

It really is a losing battle.

__________________
Hey. What kind of party is this? There's no booze and only one hooker. -- Bender (Futurama)

All of this talk about the memory bandwidth scores beating out the Athlon because of useless non-real world available bandwidth has made me thirsty.

So is it just the RAMBUS?

Maybe not! I have DDR (not RAMBUS) RAM in a P4 1.9 GHz running an i845BG and it yields the following.....

1014/1058.

An Athlon XP 1900+ with the same DDR in an AMD 761 board yields this.....

704/773.

Is the chipset that much better? Probably not. The way the CPU uses that memory? Hmmm.....

Not that those scores mean ANYTHING in a "real world", but you've got to admit that's a pretty big spread in numbers! If the same memory is used in the two PCs, where does this discrepency come from?

__________________
Hey. What kind of party is this? There's no booze and only one hooker. -- Bender (Futurama)

Sigh, thats right though. DO those numbers MEAN anything? Why dont we just say "Hey, the P4 is 2.0 Ghz and the XP is only 1.66 Ghz" OK, P4 WINS!!! We could simplify the whole proccess. My point was never which one is better. It was what benchmarks actually translate into actual perfromance on a machine, in a Game, App, Database...anything. And the Sandra memory benchmark doesnt translate anywhere to anything. And BTW, the xp 2000+ does beat the P4 2.0 in almost every REAL world benchmark (such as games and winstones). Only thing I have seen the P4 do better is real time encoding, which is where its higher number crunching actually does pay off (but still has nothing to do with its memory bandwidth). Intel makes some good chips, and features like heat throttling are way ahead of AMD, but in pure speed the AMD chips are just a small margin ahead (for comparable chips). You can't deny that anywhere.

Josh
(Running an XP2000+ unlocked AND a Piii 850)

In your article, you made no mention of what version of SiSoft's Sandra you were using? I hope it was the current (2002) version, as the older ones had some glitches...

Yes! The older version has glitches, especially in XP. It was the newest version.

__________________
Hey. What kind of party is this? There's no booze and only one hooker. -- Bender (Futurama)