PCMark 04 Record

Heres another one for that im sure im going to get slammed for posting.
<A HREF="http://service.futuremark.com/compare?pcm04=2963781 " target="_new">http://service.futuremark.com/compare?pcm04=2963781 </A>

This is the highest score in the ORB as far as I can tell.



Intel P4 550(3.4)@<font color=green>5Ghz</font color=green>
Asus P5AD2-E-Premium
Crucial Ballistix DDR2 667@<font color=red>DDR2 800<font color=red> Owned!
TT 680W PSU

Is this repeatable? Maybe you have a system bug that's speeding sh<b></b>it up?

What about 3DMark01?

In all honesty, I'm still not believing your various records....I mean c'mon, you almost doubled the score of an Athlon X2 at about the same speed as your Prescott, using very similar graphics and HDD subsystems? I'm calling shens here, again. Where the error is, intentional or not, I don't have the slightest clue.

Maxtor disgraces the six letters that make Matrox.

Sorry Rick, but no 'sweet spot' is going to give a 180%ish boost over a comparably equiped system. Something else is up here.

Maxtor disgraces the six letters that make Matrox.

The one AMD thats on the leader board is a dual processor. PCMark 04 is a weighted in the Intel direction. Im surprised that your argument didnt include the other P4's. Keep on posting all the negative crap you wish, but this will be my last reply to you.

Intel P4 550(3.4)@<font color=green>5Ghz</font color=green>
Asus P5AD2-E-Premium
Crucial Ballistix DDR2 667@<font color=red>DDR2 800<font color=red> Owned!
TT 680W PSU

I am guessing that the stopwatch refers to something said in my memory post. I havent been back in this section nor checked that post again since my last reply. The memory results are being examined so until I get an answer one way or another I didnt see the point in stressing myself out anymore by reading anymore negative replies[/Vapor]

I have split my active cooled liquid between 3 seperate pumps and tanks. Ive also done alot of work on my MOBO. I used my infrared temp gauge to find any and all hot spots on the Mobo and have made modifications to all of these places. My temps while benchmarking are within 2-3 degrees +or- 0C. Im using a mixture of deionized water and blue ice so that the feezing point changes. As you can see in my sig, ive managed to take my RAM from DDR2 667 up to DDR2 800. I seem to remember someone[/Vapor] saying that they couldnt except my memory speeds unless DDR2 800 was in the mix. Im guessing that now that I have gone past DDR2 800 that it still wont matter.

This post as well as the last one is going to catch alot of negative comments. I noticed that the memory post was up in the 700's, im sure that the numbers are much higher in the other forums that I posted the results in. Im sure by now most people have made up their minds by now regardless of what I do or say. This score doesnt have a chance of sticking for more reasons than one. The sad thing is that if it doesnt it will probably be for the wrong reasons. Benchmarking in general has alot of issues that should be addressed, in particular competitive benchmarking needs some major changes. If you look at the Hall Of Fame in the ORB scroll down on the top 20 PCMark section and look for my other score. Its around 12th. If I wanted to sneak in undetected, I would have stopped with that one and saved the bigger score for a later date. Ive got the PRO version so I could have published at anytime.

Ill make one more comment to the AMD people. If your going to discredit the scores, do it for some reason other than the one my friend Vapor has used. The benchmark itself is a terrible one to compare the two platforms (Intel and AMD) performance. The only AMD in the top 20 is a dual processor. Everyone knows that AMD is a superior platform and the fact that this is the only one in the top 20 should tell anyone that the benchmark does not show a true picture of the performance difference between the two. The first thing I did after publishing the score was to report it to the future mark team. Im an unknown in the benchmarking world so I dont expect for the score to stand. The interesting thing is going to be how they handle some of the other scores that I have found to be even more suspect than mine. One thing is for certain, if your known you dont get questioned even when some things about your setup make no sense at all when comparing them to your score. Basically its a members only club, and you better make sure your membership is good otherwise you dont even get the courtesy of having your results truley considered.


Intel P4 550(3.4)@<font color=green>5Ghz</font color=green>
Asus P5AD2-E-Premium
Crucial Ballistix DDR2 667@<font color=red>DDR2 800<font color=red> Owned!
TT 680W PSU

Crashman (who did a review for Tomshardware recently, the x800 review) would probably help you out in regards to benchmarking

-------
<b>It's a man's obligation to stick his boneration in a women's separation; this sort of penetration will increase the population of the younger generation.</b>

Oh yea, if you think that your memory speed is getting you the 3x boost.

http://service.futuremark.com/compare?pcm04=2948854

333 FSB 3-2-2-4 timing.

<A HREF="http://www.xtremesystems.org" target="_new">www.xtremesystems.org</A>

I wonder why your 14k was removed from ORB is it's legit:

http://www.futuremark.com/community/halloffame/

When did I say DDR2-800 was necessary for those speeds? I believe I said DDR2-1200 at TIGHT timings were necessary for your 11500MB/s, who knows what is needed for your 16000MB/s.

As for this score--there's no way it's real. The benchmark has favored P4s for a very good reason, a large part of the score is in multitasking, something that should be benchmarked, which A64s simply aren't good at. Now there's the X2, which will alleviate that problem for the A64s (as well as boost the other aspects of the benchmark to varying degrees).

Anyway, what wrong reasons are there for taking down your score? It's not legit in any form. Comparable systems are getting a third of your score and MUCH superior systems are getting a little more than half. Your 12th place score should also be taken down (*upon further inspection, it was taken down*) as there is no way your system can get that score legitimately.

As for me comparing Intel and AMD, whatever....your score obliterates the Intel records (which is pratically identical) comprised of faster dual-Xeon systems and significantly faster PDs. Frankly, saying that everyone knows that AMD is the superior platform is just bullsh<b></b>it, it's simply not for many tasks and if everyone did know its overall strength, don't you think they'd have the dominance in market share?

As for the members only club, go for the single P4 (non-EE) record all you want, you might be able to get it with some LN2. But the people at the top simply have the better boards, memory, and, most importantly, CPUs. Buy those parts, and you're a 'member.' You want use to consider your results? Tell us how you got them and why they should be considered even though their a good 180% larger than that of a comparable system and obliterate the WR?

Maxtor disgraces the six letters that make Matrox.

I agree with pretty much everyone in this thread
Seriously, those numbers are way too high to be valid, but something is up when more than one app reports such numbers (sandra and PCmark). Now both Sandra and PCmark are being distributed by Madonion I thought, maybe they share some code as well ?

Anyway, here is what I would do: run memtest86. Benchmarking the system at the fastest settings that will pass memtest testing might give us a better idea. If it doesn't pass memtest, God knows what these apps are reading. If it still achieves such miracle scores while passing memtest, well, then I'm curious as anyone whats up. I'd still kinda doubt you found some DDR2 holy grail of performance though, since its not likely any setup would exceed by far what is possible even in theory, but at least it gets interesting then..

Moz, can you tell us what you achieve using memtest stable settings ?

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

Trust me, if everyone <b>knew</b> that AMD were better, it would have the market dominance. Just ask the average Joe who makes the best CPUs, they'll say Intel. They also go out and buy Intel because they 'know' it's better and because they have been better for all of eternity and the bigwigs at Dell say that they'll only sell Intel because they 'know' that Intel is better.

Yes, it does seem too good to be true because it demolishes exactly equal systems and systems that are much better--did no one else bother to notice the scores of comparable systems? Did anyone else notice Mozzy's comment how those weren't his system specs? A benchmark is a benchmark for a reason: predictable inter-platform scores. Period.

And who the fu<b></b>ck do you think you are accusing me of targetting Mozz? I'd do this to anybody reporting this score with that system. Just because he has a pattern of TWO markedly-off scores, does not mean I'm targetting him. If I wanted to target him, I'd accuse him of not hitting 5GHz or not hitting DDR2-800...neither of which are particularly difficult to do, frankly. I'd accuse him of beating his children and wife (if he has them, I don't actually know). You'll know when I'm targetting someone. In fact, check the archives for me targetting SoDNighthawk...I was quite young at the time, which is why it seems sort of amateur. And as for the defeatest thinking...bullsh<b></b>it, you know just as well as I do (if you know jack about computers) that his system simply can't get 14k in PCMark04...not even half of that, frankly.

He can't be getting the various scores he's getting with his system. No one will ever legitimately get those scores with a similar setup. Frankly, a dual core PD (no difference in platforms here, so don't even try it) at the same exact clocks with faster memory and tighter timings and an R520 or G70 won't even get that score.

Am I the only one here other than Sampsa and FUGGER (who, by the way, have the top two spots in the ORB for PCMark04) that understands this?

Maxtor disgraces the six letters that make Matrox.

Im not running anything to cheat the system. I notifyied them by 2 seperate e-mails but they havent replied and probably wont


Intel P4 550(3.4)@<font color=green>5Ghz</font color=green>
Asus P5AD2-E-Premium
Crucial Ballistix DDR2 667@<font color=red>DDR2 800<font color=red> Owned!
TT 680W PSU

You sure about that? maybe it's cause they think and believe that or maybe that's what they are told or better yet taught. What Joe shmoe thinks he knows ain't always the truth. and if you really belive dell only use intel cause it's better think again. The intel inside marketing program dictates what goes inside dell as far as cpu's go.

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.

I accept PMs now...I disabled them awhile ago for who knows what and noticed it when I went to change my profile two or three days ago....should work now.

I know that the tension would have never been there if I didn't say something and alert Sampsa (who probably alerted FUGGER) of the situation. Nevertheless, I still stand by my point: something in your system (could be hardware, could be software, I really don't know) is leading to some really inaccurate scores in a variety of things.

Maxtor disgraces the six letters that make Matrox.

I'm in agreement.

Maxtor disgraces the six letters that make Matrox.

I've often wondered why (for benchmarking anyway) you don't coat the whole system in a non-conductive sealant and dunk the entire PC into freezing water or some such. Aren't there non-conductive oils and the likes that you could put into a deep freezer and then dunk the whole system when you bench?

Seriously though, I think your cooling methodology is the more interesting aspect of your OCing. You make sure that no hotspots exist. In your rig no component should ever miscommunicate because of a heat buildup. I wonder how many other OCers can say the same.

Yeah, it's possible that your scores are a bug, either in hardware or in software. (Run memtest already.) But it's also possible that your scores are a result of making sure that your system has no weak links.

<pre><font color=green><i>Jesters do oft prove prophets.</i> -Regan in
King Lear (Act V, Scene iii) by William Shakespear</font color=green></pre><p>@ 188K -> 200,000 miles or bust!

Yeah some pics should be good!

Nothing is as easy as it looks

Sorry about that vapor but your point was not very clear. I did not know quotes = sarcasm maybe that is understood here but I for one did not know that.

At least we are on the same page now that the quotes thing is cleared up:-)

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.

YOu'd also better have some odd non conductive copper

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

Thank you.

(1)SiSandra
(2)Everest
(3)PCMark04
(4)Its on my benchmarking SATA Drives cant remember but its made by the same people that put out CPU-Z

This is the same thing that I thought but its not so. In DDR2 it seems that they are still not sure what the full potential is. I will say much more on this subject later.

I want to shed some light on some things that we have all have been in the dark on. The theoretical bandwidth that Sisandra uses is not the actual formula. It was suggested to me that the reason that they use a different calcualtion for bandwidth is because they most likely are just avoiding copyright or plagerism. Im no programmer but as I understand, in the world of programming if the program didnt have a different way of calcualting max bandwidth it could be considered plaguerism.

This is where THEORETICAL bandwidth comes in. Think about what the word theoretical suggest and this makes sense. The formula they use doesnt give a true bandwidth that is why its possible to hit 100%. Ive seen the gap in the 2000 - 3000 MB/s range. Ex. One of my screen shots is around 11500MB/s and that shows up as 135% of theoretical bandwidth. The actual max bandwidth would be closer to 13000MB/s

Ive run memtest 86 but im not confident that I did what I shopuld have done. It passed this test, but if someone could fill me in on how to use the program I will run it however I need to. I dont have a problem running the others, and will do so. My online time has been cut in half so that combined with the stress I had let all of this turn into are the reasons that you see me post and then not return for a couple of days. I think that you are on the correct path in most of what you have said. I will download the other programs you mentioned and get to work on them. Let me remind all of you that I am in the process of having a third party research validate(or find the glitch) my results and I promise everyone that I will disclose everything. I wish I had more time to tell you all about some things that have been on mind in regards to the FutureMark benchies but it will have to wait until later.

I was in a hurry, no time to read back over this. Hope it made sense.



Intel P4 550(3.4)@<font color=green>5Ghz</font color=green>
Asus P5AD2-E-Premium
Crucial Ballistix DDR2 667@<font color=red>DDR2 800<font color=red> Owned!
TT 680W PSU

Understood. I will try and make it a point to PM you in the future if I have something on my mind that doesnt belong in the open forum. Thanks.

Intel P4 550(3.4)@<font color=green>5Ghz</font color=green>
Asus P5AD2-E-Premium
Crucial Ballistix DDR2 667@<font color=red>DDR2 800<font color=red> Owned!
TT 680W PSU

>(1)SiSandra
>(2)Everest
>(3)PCMark04
>(4)Its on my benchmarking SATA Drives cant remember but its
>made by the same people that put out CPU-Z

I dont know everest, or the unnamed program, but I guess that pretty much rules out a software bug.

>This is the same thing that I thought but its not so. In
>DDR2 it seems that they are still not sure what the full
>potential is. I will say much more on this subject later.

Nonsense. DDR2 is designed, specced, documented.. what you are saying sounds as likely as AMD or Intel not knowing that enabling L2 cache in the bios would give a big speedup :|

What they might not know exactly yet is how far it can clock, or at what timings, since those are manufacturing related, but the design of these things is completely utterly known. Further more, the performance is known, don't you think that was tested, even simulated while defining the specs ? Don't you think chipset vendors and bios developpers know what they are doing ? "hey we built something new here, but we're not quite sure what its capable off".

Whatever BIOS setting you used, they where put in there by a bios developper for a reason, and by following guidelines written by the guys that designed the chipset. Really, no offense, but there is nothing magical you (or me, or anyone) could discover that would not be *known* by the guys that designed it. certainly nothing boosting performance two fold or whatever it was (your link was removed).

Seriously.. *think* for a second.. you honestly believe you discovered something, that would make intel turn its head and say "hey, look at what Mozzy did, lets implement that in our Bios and chipsets and we'll get >50% better memory performance at a given clock and DDR2 latency !".

>Ive seen the gap in the 2000 - 3000 MB/s range. Ex. One of
>my screen shots is around 11500MB/s and that shows up as
>135% of theoretical bandwidth. The actual max bandwidth
>would be closer to 13000MB/s

I'll rely on others to check your math, I don't have your numbers here.

>Ive run memtest 86 but im not confident that I did what I
>shopuld have done. It passed this test, but if someone could
> fill me in on how to use the program I will run it however
>I need to.

There are not many ways to run it. Just make the bootdisk, boot from it, let it run a full test for a few hours. If indeed it passed that at the frequencies and settings you benched your computer, then indeed I am very surprised.

<b>Can someone please as kind as to summarize for me, the bandwith figures Mozzy achieved, what others get with similar clocks, and what the theoretical bandwith would be for such frequencies ? I was under the impression you doubled or tripled bandwith scores of otherwise comparable systems, now I am confused. Either other setups would be unbelievably inefficient for some reason, OR you exceeded theoretical bandwith (which is impossible) OR the performance gains you achieved are much more modest than what I thought. </b>

BTW, as I recall memtest86 also reports memory bandwith figures (top left of the screen). It would be extremely interesting if you could post those as well.

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

There are two worries about using other people's code in software:

1) Copyright. In the US at least, technically all text is copyrighted. So if someone copies your text verbatum without permission then you can (theoretically) get in trouble for that. Of course there are fair use laws that make exceptions to this, terms and agreements of service usage that may grant permission without you realizing it, and, as always, the copyright protection itself only protects so long as a certain percentage remains unchanged. So I could take someone else's code, rename the variables or make some other significant change, and now it's no longer under copyright protection.

2) Patent. Patents generally cover a method of doing something instead of a specific section of text. In fact many patents don't even have a working example because it's the method that matters, not the implimentation of that method. So any patented algorithm cannot (legally) be reproduced without permission. Now, there is however always more than one way to skin a cat, so just because someone has a patent on an algorithm doesn't mean that you can't use that as a basis to determine your own (slightly) different method to get the same results. Plus many times patents have a very specific scope of their usage, so using it in a manner outside of that scope is also legal.

So even if the code was copyrighted, that's no big deal for someone willing to change that code's appearance significantly. But if it was patented and they don't want to license the patent, then yeah, they probably developed their own algorithm.

And when words like theoretical or approximate are used, one must always take the accuracy with a grain of salt anyway.

<pre><font color=green><i>Jesters do oft prove prophets.</i> -Regan in
King Lear (Act V, Scene iii) by William Shakespear</font color=green></pre><p>@ 188K -> 200,000 miles or bust!

You read way to far into what I said. All I was really talking about was the speed and the timings, not the design.

I know from talking with reps from more than one RAM company that they dont even test their RAM with liquid much less anything colder than that. The explanation that I was given is that the market just isnt that large for people who use liquid. I was under the impression that they would probably make every effort to test their RAM with liquid at the very least, but the ones that I have talked to do not. I will send pics in the next couple of hours.

Intel P4 550(3.4)@<font color=green>5Ghz</font color=green>
Asus P5AD2-E-Premium
Crucial Ballistix DDR2 667@<font color=red>DDR2 800<font color=red> Owned!
TT 680W PSU