SICK SCORES!

P4Man · May 13, 2005

>Im going to try to get something a little more organized on
>my website. Ill post a link shortly.

Please do.

>As for ORDINARY, as new as DDR2 is im not so sure that
>ordinary is a good word to use at this point.

You saw the context I put that in. There is huge difference between something just being fairly new like DDR2-533(+), and setting world records among LN2 cooling, voltmodding soldering, 3 CPU's and 2 MBs per week frying, diehard overclockers a la Fugger that often even get cherry picked parts from manufacters.

Somehow, I doubt you beat those guys, but I'm keeping an open mind.

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

endyen · May 13, 2005

A 5ghz chip aint something too average, no?

P4Man · May 13, 2005

>A 5ghz chip aint something too average, no?

Its considerably more than what i've seen myself, but its also no more than a clockspeed achieved by THG in 2003, and considerably less than the 7 GHz or so Fugger achieved. For normal people, there is just no competing with them.

So is 5 GHz a great overclock ? yes. Anything near a worldrecord ? Nope. I would assume his memory overclock/performance scores are something similar.

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

mozzartusm · May 13, 2005

Im curious as to what % of the ones that have hit 5Ghz did it with liquid. Im thinking a small portion. I think that many people get excited about the next level being broken and then when it has been its not a big deal anymore. The truth is that its still a very difficult task especially when its done with parts that are bought in the normal consumer market. Ive been chatting back and forth with some guys in other forums about getting their systems to OC higher. The questions are almost always the same. "What settings are you using?" Ive asked and probably will ask the same questions. Knowing what FSB and how much voltage will get you a long way, however after a certain point it comes down to fine tuning each individual system according to the circumstances. Last night I was reminded of this as I noticed that the lower the temps outside dropped the higher my rig would OC. More signigicant than that however was how much smoother it ran.

Im still blown away at the fact that nobody in the open forum asked me for proof when I posted that I broke 5Ghz. A couple of people asked later in PMs but no challenges in the open forum. That was a very difficult goal for me to obtain and wouldnt be easy if I were to try again. The memory speeds were hit as a result of trying to break 5Ghz. I took the first pics back in Jan of the same speeds that I have recently posted. I knew then that they were very high and decided to not do anything with them. The way it happened makes me wonder if things would have been different had I posted them back then. I think things would have been very different. I decided that instead of jumping back and forth between settings that I would apply some of the research knowledge that I have been exposed to in college. I put together a research design that would help me to systematically reach the highest clock speed that my system would reach. I began to isolate three parts of the system so that I could find what each part was capable of reaching. First thing was to drop the multiplier from 17 down to 14. This brought the CPU clock from 3.4 to 2.8 Thats 17 X (FSB)200 = 3.4 down to 14 X (FSB)200 = 2.8. This along with dropping the memory to a low speed and loosening the timmings, gave me a chance to see how far the MOBO was capable of OCing. When I found the max FSB for the mobo then I would know later on if it was the component that was the limiting factor. Next I tested the RAM by raing the Multiplier back up to 17 and using a divider to OC the RAM. To my surprise, the RAM was far more capable than I had give it credit to be. So now I know the limits of my Mobo and the limits of my RAM. There was no doubt in my mind that the CPU would be the limiting factor. Boy was I surprised at the outcome. I still dont know what my CPU could do given the right circumstances and a properly Modded Mobo. At 5Ghz the CPU was not what kept me from OCing higher.

What I was attempting to point out was that if I would have reported my memory speeds as I climbed up with my OC would this have been accepted differently. This is an idea of what you would have seen. The numbers are rounded off. This was obtained from screen shots that i have from the aprox time of year and months.

Dec 1st
4800 MB/S
5100 MB/s
5300 MB/s
5600 MB/s
6200 MB/s

Jan 1st
6800 MB/s
7200 MB/s
7400 MB/s
7800 MB/s
8400 MB/s

As you see, the picture would have looked much different than it does now. Over this period of time everyone would have seen the steady progression from 4800 to the 8000 MB/s mark. Would this have been seen in a different light if the screen shots would have come in over a three month period and gradually increased as I gradually clocked my rig higher?

Intel P4 550(3.4)@5Ghz
Asus P5AD2-E-Premium
Crucial Ballistix PC2 5300 2X256
TT 680W PSU

mozzartusm · May 13, 2005

Nope. I would assume his memory overclock/performance scores are something similar.

My best guess right now is that I have done something a bit different than some of the others that have hit these high OC's. Without knowing the BIOS settings that each one of them used there is know way for me to know for sure if anyone has tried the exact config that I have been using. From reading over the threads that I know of, there does seem to be a couple of differences one of which I have yet to sse anything written about. Im going to post the details in a few days but because of the speculation at this point im waiting until I see if I can get this verified through a third party. If im wrong then so be it, I did what I thought was the right thing and reported the numbers truthfully. I hope something positive will come out of this no matter what we find to be correct or incorrect with the benchmarks. My goal is a simple one, and that is to do everything that I can to push the industry towards new and improved toys. This is part of a reply that I made in another forum. Ive been wanting to post it, the second part sums up my intentions.

Originally Posted by mozzartusm
Hey guys! First I want to thank all of you for the postive feedback. I am new to this forum so I didnt know how you would react. Ive posted my scores in more than 1 location and 95% of the replies have been positive, however there are a few people that have outright accused me of posting fake scores. This really bothers me, and even though I know that I shouldnt let this get to me, it still does. If anyone has any suggestions as to how I could further validate or prove that I am not doing anything to alter the scores then please post your suggestions. I would like to get rid of any doubt that people may have. My goal is a simple one, I am dedicated to pushing the industry anyway that i possibly can. I think that most of us want the same thing, and thats simply newer, better, and FASTER technology. The online communtiy and especially Overclockers/Computer enthusiast do an excellent job of pushing the edge and keeping competition among manufactures high at all times. This is probably one of the biggest factors that has lead to us to where we are today. It wasnt that long ago that 500Mhz was blazing fast for a CPU. Any suggestions are welcome!

Thanks

Mozz

Intel P4 550(3.4)@5Ghz
Asus P5AD2-E-Premium
Crucial Ballistix PC2 5300 2X256
TT 680W PSU

mozzartusm · May 13, 2005

I may have looked at the wrong numbers but isnt it around 12000 MB/s

Intel P4 550(3.4)@5Ghz
Asus P5AD2-E-Premium
Crucial Ballistix PC2 5300 2X256
TT 680W PSU

Vapor · May 13, 2005

WR is 11400 or so with dual Optys w/ NUMA enabled, IIRC.

No Intel system can reach that number yet....maybe when we see DDR2-1200 with 3-2-2-4 timings, MAYBE.

Maxtor disgraces the six letters that make Matrox.

Rugger · May 14, 2005

I'd be extremely interested in seeing your detailed setup. Especially what settings/tweaks that you are making that have been overlooked by everyone else.

Are you posting the details Mon?

__________________________________________________
You're a boil on the arse of progress - don't make me squeeze you!

Vapor · May 14, 2005

No matter what the progression was to those numbers, they aren't real! For one, it's impossible to have 116% efficiency. Second, there's no way you'd even have the Intel WR with DDR2-714 with so-so timings. I mean, there are people out there at DDR2-800 with 3-2-2-4 timings and still aren't even close to your numbers.

Maxtor disgraces the six letters that make Matrox.

Rugger · May 14, 2005

Just because other people getting higher OCs on the memoyr doesn't mean that their OCs are as stable or tweaked any better than the setup that Mozz has running. There's a difference in posting an OC and getting a stable/fast OC. Rarely do you see people running their rigs at their highest OC, because the highest OC is rarely where they get their top performance.

You still haven't responded to the similarities in scores across multiple bench utilities over an extended period of time. Just because you find something hard to believe or impossible doesn't mean that they aren't real. If that were the case, then the world would still be flat. :/

__________________________________________________
You're a boil on the arse of progress - don't make me squeeze you!

Vapor · May 14, 2005

Just because other people getting higher OCs on the memoyr doesn't mean that their OCs are as stable or tweaked any better than the setup that Mozz has running. There's a difference in posting an OC and getting a stable/fast OC. Rarely do you see people running their rigs at their highest OC, because the highest OC is rarely where they get their top performance.

And? What's your point? He could post his highest OC and it would still be wrong. Stability was never a point I brought up because no one ever breaks a WR with their 24/7 setup. Needless to say, there have been some recent advances in DDR2 technology that enables people's 24/7 stable settings to be faster than what his memory is capable of doing at its peak. Also, It's safe to say that Mozz isn't the best tweaker out there. Most, if not all the top 5, are at XS running DDR2 at significantly faster speeds with tighter timings and aren't getting the same numbers. The Japanese aren't getting these numbers either, and they're just as good as the guys at XS.

You still haven't responded to the similarities in scores across multiple bench utilities over an extended period of time. Just because you find something hard to believe or impossible doesn't mean that they aren't real. If that were the case, then the world would still be flat. :/

It's a bug, it's a proven system bug. He is getting those numbers but in no way do they resemble what his system is ACTUALLY capable of doing. If he runs a non-synthetic benchmark that is almost entirely dependent on bandwidth or even memory performance in general, he will get the same performance as someone without the bug at the same settings. No one has a way to fix the bug yet as it's been somewhat random. The easiest way to tell who has it, however, is to look at the Sandra efficiency. >90%, it's a bug. 85-90%, speculative bug. DDR2 isn't efficient yet (might not ever be), it has a LOT of theoretical bandwidth, A LOT, but its timings and other intricacies hold it back (and it always will be like that--DDR can barely reach 90% efficiency). Mozz's efficiency is at 116%--does NO ONE ELSE FIND THIS INTERESTING????? It's a giant red flag that the results are buggy, across all bandwidth-measuring apps. When he runs it, does he notice his FSB fluctating? He should as it's another tell-tale sign of the fact that it's a buggy result. There's nothing wrong with the programs, it's Intel's throttling or something. It changes the system clock mid-test, which causes an error in all bandwidth-measuring tests that is displayed as an egregiously large number.

Am I the only one that truly understands this stuff??? No doubt Mozz has a good system, a great deal better than most others here, but the numbers he is getting simply aren't the truth.

Maxtor disgraces the six letters that make Matrox.

mozzartusm · May 15, 2005

Hey Vapor,

I think that I have heard you loud and clear. No need to continue saying the scores arent real. You have brought up some points that I have and am currently researching. Theres really no need in reminding everyone that this is not legit. I assure you that I know where you stand. As for the 116% being impossible. Im certainly not an authority on this subject, but I have been talking back and forth with someone that is an authority. Im sure that you have heard of Crucial. Im not going into detail"I doubt you want to hear it anyhow" but we are working on getting the results validated or finding what is causing them to be incorrect. LET ME BE CLEAR SO THAT THIS DOESNT GET TURNED AROUND. THEY ARE GOING TO LOOK INTO THIS AND ATTEMPT TO VALIDATE THE SCORES IF THEY ARE CORRECT OR INVALIDATE THEM IF THEY ARE NOT CORRECT. This is a bit tricky being that I live so far away from their Lab. So if I can get things on my end situated so that they can see first hand then this should settle things one way or another. If you will just save the negative "fake" comments until they say one way or another then we will know for sure. I never expected everyone to take my results as the final say so. I also never expected to get slung around in the very forum that I call home. Most of my friends here have been very supportive even though some of them as well as myself have some doubt as to how this will turn out. I would not expect for anyone to take up for me just because this is home, however many of them have and I am thankful for that. This has been a rough experience as I am trying to keep my head above water in more than one place at one time. I have gotten alot of e-mails, but one in particular stood out. This person wrote "Hey Mozz dont let them get you down. I would have done the same thing if those benchmarks were mine. I know that not everyone here "Here=his forum not THG" is easy to get along with but once they get to know you they will lighten up. If you get tired of their "PEEP" you are welcome here.

He was talking about THG. I wrote him back and told him that it wasnt as bad as he thought it was and that he should read through it again. he replied once more and said this. "I see what your saying. I was so caught up with the guy that was saying your scores were fake that I didnt notice the good stuff."

Im not sure what your intentions have been, but from other peoples point of view they see your "FAKE SCORES" subject lines and automatically think that everyone in my on forum is down on me worse than in the ones where im not known. One way or another, this is gong to be validated or found to be errant. Im sure you prefer that it be the latter, and if so then you can kick me while im down. Keep the faith my friend. You may get what your hoping for.

BTW, if you want the formula that shows how theoretical bandwidth is determined I can send it to you. I hope you have better luck than me in understanding it. 100% isnt a brick wall you know.

Intel P4 550(3.4)@5Ghz
Asus P5AD2-E-Premium
Crucial Ballistix PC2 5300 2X256
TT 680W PSU

mozzartusm · May 15, 2005

If he runs a non-synthetic benchmark that is almost entirely dependent on bandwidth or even memory performance in general, he will get the same performance as someone without the bug at the same settings.

Ive been asking all along for any suggestions so if you know of other programs that I should run then please post them.

Intel P4 550(3.4)@5Ghz
Asus P5AD2-E-Premium
Crucial Ballistix PC2 5300 2X256
TT 680W PSU

mozzartusm · May 15, 2005

What? Me push him? Hes been posting "fake" from the beginning. Im confused.

Intel P4 550(3.4)@5Ghz
Asus P5AD2-E-Premium
Crucial Ballistix PC2 5300 2X256
TT 680W PSU

Vapor · May 15, 2005

Mozz,
Your scores aren't real--I don't need a professional to tell me this, there is a known bug with SOMETHING (it happens in all measuring proggies, just SPI, WinRAR, etc. don't reflect the added bandwidth). I'm not trying to put you down...but everyone jumps the gun on the kudos so quickly and they don't sit back and look at the numbers.

116% efficiency isn't possible, as I've said. Take your FSB (1171 I think it was...), multiply it by 8, that's theoretical. The reason why 100% is impossible (and so is 90% with DDR2) is latency. Look at <A HREF="http://www.xtremesystems.org/forums/showthread.php?p=862328#post862328" target="_new">this</A> OC at XS, 1360FSB, DDR2-906, 4-2-2-x (tras isn't that important) and he gets 8500 buffered, with 78% efficiency. The Stilt is one of the best OCers in the world, using faster, lower latency RAM than you getting a lower score and a believable efficiency.

I'll also reference <A HREF="http://www.xtremesystems.org/forums/showthread.php?t=58874" target="_new">this</A> thread at XS, other people are getting the same bug. Run CPU-Z while you test bandwidth....notice something fluctuating? Thought so. Not sure if it's mobo makers' trick or Intel's--but it tricks the proggy into using a different reference value for the system clock essentially.

Look Mozz, I'm not trying to be an asshole, but wake up and smell the coffee. Do some research, look around at some more knowledgable places than here and OC.net...notice that you're not the only one with weird results. Notice that most people have figured this shit out, though....it's a bug/trick/something. The scores aren't real, run SPI 1M....with a 16000MB/s buffered bandwidth (as per one of your screenies at overclock.net), you should break 20 secs, or at least own for the ~4GHz category....but it won't.

Sorry Mozz, I know you asked me not to respond like this...but I'm just letting you know what Crucial is going to say, or what some searching would come up with. Just putting the facts out in front of you and everyone else and shedding some light on the truth.

Maxtor disgraces the six letters that make Matrox.

Vapor · May 15, 2005

Yeah, RUGGER did kinda piss me off. Whatever....I voiced my final statement on the situation until Crucial says something.

Maxtor disgraces the six letters that make Matrox.

Rugger · May 15, 2005

I'll throw in my final word for now, too. My previous was more to defend Mozz than to piss you off. Your posts were accusatory/demeaning in nature and that kinda lit a fire under me, too. I apologize if I ruffled your feathers hard.

And? What's your point?

A couple of points that I was trying to make:
1. Just because others have not been able to do something doesn't mean that the task is impossible or that Mozz can't do it.
2. Just because a person has one of the current WRs doesn't mean they have a monopoly on new/innovative approaches to an OC. At some point the WR holders were just enthusiasts like you, Mozz and I.
3. I brought up stability because some OCs are claimed when you see the numbers in the BIOS. There is difference between in being able to get into the BIOS to claim an OC and booting the system/running benchmarks. Additionally, because a system is running at higher speeds doesn't mandate that they will have higher scores in benchmarks. If the system cannot access the memory stably, then it may not score as high as a system that is at lower settings but significantly more stable.

I completely understand your points IRT efficiency and random system bugs - I'm not refuting them. What I am choosing to do is stand with Mozz. If the scores are proved to be inaccurate because of a system bug in the software, then so be it. If they are proven to be accurate, then so be it. Either way I'll stand with him.

__________________________________________________
You're a boil on the arse of progress - don't make me squeeze you!

Vapor · May 15, 2005

I did come out hot and accusatory, no doubt about it--I'd edit my posts to fix it, but there's a bunch and I'm lazy :tongue: . Everyone knows where I stand on those topics anyway...

I gotta give Mozz credit, he is a good OCer, better than I am (if for no other reason than I bought a mobo and RAM combo that just won't OC well). As good as he or anybody else is....there isn't any Intel hardware that can truly produce the numbers he is getting, maybe at DDR2-1200 or so (with fairly good timings [4-3-2-x, maybe]).

As for point #3, I agree. BIOS shots are LAME. Memtest shots are fairly acceptable under circumstance (when FUGGER got 7.2GHz, but there was no SetFSB or Clockgen for his board/chipset, for example). The best shot is a repeatable benchmark shot (or fairly repeatable...for WRs, I'd take repeatable to be: can be done one out of fifty times successfully, but w/e. While true stability does help scores (3DMark, for example), there are DDR2 modules out right now that can run at a higher speed and tighter timings that what Mozz has, guaranteed out of the box (OCZ 800EB, for example), that get lower scores (but the best scores out of all DDR2).

Anyway, that's all for now.

Maxtor disgraces the six letters that make Matrox.

slvr_phoenix · May 16, 2005

Mozz's efficiency is at 116%--does NO ONE ELSE FIND THIS INTERESTING?????

As I already said in this thread, "several numbers that Sanrda uses are approximated. So it's no wonder you're over 100% efficiency. It is by defenition less than perfect math to begin with."

Did I notice it? Yes. Did I find it interesting? Only marginally so. If Sandra didn't use so many approximations then maybe it would be a little more interesting. As it is, it just represents a bug in Sandra's approximations.

<pre>Jesters do oft prove prophets. -Regan in
King Lear (Act V, Scene iii) by William Shakespear</pre>@ 187K -> 200,000 miles or bust!

Spajky · May 16, 2005

Memory efficiency & WinRAR´s built_in test benchmarking memory: see my research article under comp/benches -updated on my site ... 🙂 ...

--
Regards , SPAJKY ® http://www.spajky.vze.com
3rd Ann.: - "Tualatin OC-ed/BX-Slot1/inaudible setup!"

Spajky · May 17, 2005

>Is this like the new breed of overclockers?<

No, actually is a very old one; started changing quartz cristalls on ZX81 (Sinclair) board "computer" 🙂 something about a quarter of century ago; could be that you were not even born yet at that time .. 🙂)

PS.: Nice site you have .. 🙂

--
Regards , SPAJKY ® http://www.spajky.vze.com
3rd Ann.: - "Tualatin OC-ed/BX-Slot1/inaudible setup!"

ChipDeath · May 17, 2005

started changing quartz cristalls on ZX81 (Sinclair) board "computer" 🙂 something about a quarter of century ago; could be that you were not even born yet at that time .

Hey! I always wanted to do that to my 48K Spectrum... Or rather I wanted to replace the Z80A (could allegedly run @ 4Mhz) with a Z80B (6Mhz). My brother told me at the time I'd need to change the crystal too, but I never really got around to trying anything. I didn't really understand that much of that stuff back then.

What was an Overclocked ZX81 like then?

---
"Life is not like a box of chocolates. It's more like a jar of jalapeńos - what you do today might burn your ass tommorrow."

Spajky · May 18, 2005

IIRC Refreshing TV screen was less jerky; w/ some problems with picture synchronization on TV w/ 4Mhz Q, IMHO the original was less than that; all was a Kit (cheaper) I bought with a friend (these days programmer!) & we soldered stuff together, Cpu or some other chip was not a good exemplar ... LOL Sinclair Basic .. 🙂 ... oh, those were the times & quite expensive hobby than for me ...

--
Regards , SPAJKY ® http://www.spajky.vze.com
3rd Ann.: - "Tualatin OC-ed/BX-Slot1/inaudible setup!"

P4Man · May 18, 2005

Hey mozzy,

A few words from uncle p4.

Vapor is right. As I see it, this is as much a WR as my 37 second lap time on my SV1000 is on the Nuremberg ring.... using a broken stopwatch. My bike is still not quite a slouch though, and I'm not all *that* terrible behind the controles, just not on par with pro's or semi pro's on their LN injected hayabusa's or factory bikes.. and I don't need TAG Heuer to confirm that for me either.

>One way or another, this is gong to be validated or found to
> be errant. Im sure you prefer that it be the latter, and if
> so then you can kick me while im down. Keep the faith my
>friend. You may get what your hoping for.

Oh please.. don't get so emotionally involved in what a completely stupid program like sandra says in something as utterly irrelevant (in the overall scheme of things) as how many bits get transferred per second from one commercially available piece of melted sand to another... not to mention what anonymous posters, myself included, have to say on the subject.

Once your hand has healed, try some fishing...

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

lb19984 · May 19, 2005

wow what a post.......
WAit are there pics up of ure rig Mozzartusm ? I looked around and i couldnt find a link anywhere
thanks
Very impressive benchies and OC by the way :-0

Mozzartusm i also sent u a private message

<A HREF="http://www.freewebs.com/lb19984/" target="_new">My Rig</A>

SICK SCORES!

Distinguished

Splendid

Distinguished

Splendid

Splendid

Splendid

Distinguished

Splendid

Distinguished

Splendid

Distinguished

Splendid

Splendid

Splendid

Distinguished

Distinguished

Splendid

Distinguished

Splendid

Distinguished

Distinguished

Splendid

Distinguished

Distinguished

Distinguished

Share this page