The 9262 vs XPS1730 3Dmark06 score fiasco

Well finally I have a little bit of time to add my 2 cents to the battle raging between the XPS and the 9262 3Dmark 06 score fiasco.

Let us look at it from an unbiased point of view and look at a couple of independent reviews I found on the net regarding the XPS M1730 first, then we’ll analyze the data.

Here are 2 links to reviews of the XPS 1730:

http://reviews.digitaltrends.com/review5036_main25188_page2.html on this page you will see under Performance that the XPS scored 12,704 in Futuremark’s 3DMark06.

http://www.extremetech.com/article2/0,1697,2260219,00.asp on this page the XPS scored 12717 in 3DMark06. Also please read the small paragraph on top of the 3Dmark06 graph which pretty much states that faster CPUs do make a difference in 3Dmark06 scores.

Now let us do some math. According to this post here http://forum.notebookreview.com/showthread.php?t=234162&page=2 , this post here http://forum.notebookreview.com/showthread.php?t=236240 and my own 7950GTX SLI the increase in 3Dmark06 an SLI setup will give over a single card setup is about 3600 (SLI Score Minus Single Card Score) points.

Now if we look at the scores, according to the review above we will conclude that a single card on an XPS system will give us a score of 9117. The SLI score minus the difference between sli and single setup (12717 minus 3600). Which is exactly where a Core 2 Extreme mobile CPU at 2.8Ghz score would be.

By comparison, if we look at the post here http://forum.notebookreview.com/showthread.php?t=236240 by Swiftnc we will see that the 3Dmark 06 scores both in SLI and single are about 860 points higher than that of the XPS. That is where his scores should be given the fact that he has a Quad Core Q6700.

Looking at the few facts I have gathered in a short time and according to my analysis of the data (of course I could be way off in my analysis, doubt it thought..lol) I think, and that is my opinion, that the 8800GTX cards used in Clevo and in Dell are very comparable. I think they perform very close to each other without either one outperforming the other. The difference in my opinion will be negligible.

Now if we want to make a more in depth analysis and comparison of the 2 we need to look at the SM2.0 and SM3.0 scores independent of processor. I am not implying anything by what I am about to say, I am merely stating the facts as I see them. Now I do not know what the stock speed for the 8800GTX are but according to the post here http://forum.notebookreview.com/showthread.php?t=234162&page=10 by duane123 his SM2.0 and 3.0 scores do not add up (at least in my opinion) unless the GPU is overclcoked. I realize that he was running the CPU at stock speed but what about the GPU? Again I am not implying anything I am merely thinking out loud.

Maybe someone or duane123 could run a test at stock speeds for CPU and GPUs.

At any rate in real world operations I think that the 9262 will out perform the XPS for the simple fact that it runs a Desktop CPU at 1066mhz FSB and at 1333mhz FSB, also it runs memory at 800mhz not to mention quad core.

Now both laptops are great, they both have excellent performance all around. Think of them as 2 Supercars with one, being the NP9262, of them a bit faster in the quarter mile due to the extra horsepower it has under the hood, i.e the higher FSB.

Cheers
The WackMan

 

"I think they perform very close to each other without either one outperforming the other. The difference in my opinion will be negligible."
What you're trying to say is that there is no performance difference, otherwise one would outperform another. I also think that some scores on the net don't actually reflect real-life situations and that for instance, the first tests that were made on the M1730 actually showed a very minor (less than 3k) difference in 3DMark 06 (take a look at the NotebookCheck review). What should people make of this?

"
Now both laptops are great, they both have excellent performance all around. Think of them as 2 Supercars with one, being the NP9262, of them a bit faster in the quarter mile due to the extra horsepower it has under the hood, i.e the higher FSB.

Cheers
The WackMan"

Both fat, both ugly, both fast...which supercars ?

 

Now Now Fade, on your remark about them being Fat, Ugly and Fast… You forgot to mention that they are BOTH FREAKING HEAVY…lol

 

While they may perform similarly, isnt the NP9262 cheaper than the XPS1730 when equiped with similar configurations?

 

I put so much into this stuff bud.

My card is overclocked and can't beat him in any scores single vs single I can match him in total score but my CPU score is 700 points higher than his.

we both are overclocked the same and using the same drivers.

If you take away 700 points from my score and then we would have the same processor score I would be under by quite a bit.

and btw my sm2.0 scores never ever are as good as his single vs single same drivers and overclocked.

I am the one testing agianst and the one who brought this stuff to light.

My card in the tests is overlclocked the same as duanes who helped me by comparing and providing results .

it's nice to post a whole bunch of links but I have actually put down the time to test. and duane as well.

If you read the 8800 sli thread. there are tests in there that contribute to all I have been saying for the past 2 weeks.

with my tests equal to his in drivers and overclock.

I have allready proved in realworld tests between duanes, magnuses and mine.

that Clevo cards are underperforming compared to dells.

the systems we are comparing is

single m1730 174.31 drivers 600/950 clocks x9000

to mine with same drivers and clocks with q6600

I have allready posted nearly 100 posts on this stuff.

and proved it countless times.

we have also proved that overclocking the x9000 has not that much to add to the scores...

 

Yes, but fat = heavy. I can't be around 260 pounds without being both...

 

do you guys think a m1730 with sli 8700m gt for $1500 will be a good buy?

 

Unless your really tall?

 

How do your stock speeds compare on the GPU Dexgo?

 

Oh and Dex, I understand where you are coming from but you have to realize two things.

1. Dell is a huge vendor, they have more money so they are working along with Nvidia to release a better card. It is sad but true that Sager as it is not well known probably got a back burner BIOS and thats it. Now it would be nice if the cards were the same, I know I wish they were (and thought they were) when I ordered my Sager, But still I would never order a Dell just because of a 5-10% difference In video cards. Plus the Sager has a more powerful CPU - which I need more then a few extra points in a benchmark utility that proves nothing.

2. The cards are still very new, there will be updates that should even us out in the end. I saw it before with my Desktop SLI rig. When I first got my cards I only got a 30% increase in SLI. Now with new drivers I see 85-90% increase.

Still with the better processor we are on par with Dell, at an (average) lower cost. Things are not as bad as you make them out to be.

Edit** I swear if someone says (well i paid xxxx.xx for a dell in X country where the Sager would be xxxx.xx here) I'm going to shoot you over and over again (pick the game). People get deals on both systems. Some pay less some pay more. In the US the Dell vs. Sager is a good 700 to 1,000 dollar difference w/o haggling/knowing someone/coupons/or whatever else you can come up with.

 

psycroptik, I'm 6' 2.5", but...

 

I agree completely!

I did a search for "3DMark06" in the Dell 1730 Owners Lounge and looked at scores for the stock version of the SLI machine. The average for the first six scores I found was 11,900 at the default resolution.

Swiftnc has just posted a score of 13003 for his SLI 9262. The difference is undoubtedly due to his Q6700 CPU.

 

right. and I was comparing to an stock x9000 wich gives you 13k as well with stock settings.

their best system is x9000 with doesn't compare to our system with q6600

 

read the sli 8800 thread. I posted everything in there.

 

Which is 2.8Ghz as compared to his 2.66Ghz.
Dexgo Have you ever added up the 3 scores from 3dmark06 and notice that they dont equal the total number? So your theory of -700 this and +1000 that wont work. Its not a linear test.

 

his score is lower than mine.

3000 CPU score. 2.8ghz

3700 CPU score 2.4ghz


I know it doesn't add up.

but the sm2.0 scores are still lower.

and 3.0 stock/overclocked. single vs slingle sli vs sli.

I am so done with this junk

I'm tired of it.

/thread

 

Well, just in case anyone cares, after my underwhelming performance on t'other thread , I think I'll sit this one out - anyone for ?

 

Shyster! Can I get some?

 

psycroptik when you get yours.. then you can jump in and compare...

but I allready know the outcome.

I know the limits of our gpu's.

and they won't overclock without the SHADERS being UNLOCKED. or a certian bios rev to allow on the PC as well.

ALSO we need a 1.05 voltmod just to get over 1350 shaders and 900 mem.

ours can't even go over 600 CORE!!! no matter what.

Compare this stuff to the m1730... wich has no problems overclocking with software and going way over these without any voltmods or bios hacking etc etc....

also the m1730 runs fine way higher clocks than the clevo's with 1.00volts..

I know the limits of these cards inside out... trust me on that atleast..

 

So your whole new hatred for the 9262 stems off the score given by 3dmark06 (specifically only the sm2.0 and sm3.0 scores) even when we have higher overall 3dmark scores and what seems like similar/better fps at a lower cost overall?

 

Oh look, another thread about this.That`s it, I`m getting an AW m17x SLI 8800M GTX and will pwn all of you!

 

*sigh*

Finding random 3dmark scores and saying "Look see it's lower!" doesn't mean anything. Me and dex have been doing controlled tests. The same drivers, the same clocks, etc.

I'd like to state that my goal here isn't to tout how much bigger my epeen is or to tell everyone to buy the dell. I am simply trying to help Dex and others figure out what's going on. Perhaps it's something that can be fixed in bios, perhaps it's just a fundamental difference in manufacturing.

Saying the scores don't add up doesn't invalidate the fact that the scores are lower. We know that he is getting a lot more points from CPU whether it's 1000 or not it's significant.

The goal is to figure out why it is this way, the "if" has already been answered.

And for those asking I got 14290 running no GPU overclocking processor at 3.4 using 174.31 drivers with SLI enabled. I can post screen caps later if it's importat.

 

On it's way ... soon as I can figure out how to jam it into this little slot thingy on the computer

 

In some of the other threads some were comparing a OC'd 1730 to a stock 9262. I agree with psycroptik, that we should be comparing real world use (as you have always touted Dexgo) instead of 3D Mark 06. It is also strange; as I previously reported and as psycroptik stated above, that the scores don't add up to the total 3D Mark 06 Score.

Since I was unable to post in the other thread due to it being closed, here are some tests that were we just completed:



However it is now a even bigger mystery. CPU score with the 1333FSB has its advantages (Run 4) as it got a little better score then the [email protected], but not better then [email protected] setting.

Notice how the CPU score received a big jump on Run 5 but it did not reflect the same big jump in the “3dmark score” while it did improve. But if you look at the Sm2.0 Score: and the HDR/SM3.0 Score on the Q9550 it is even worse then the E8400. So it looks like more cores does not out run the Ghz differences here.

Notice how the score on Run 6 is very close to the Run 2 setting NP5793, with the advantages of faster FSB.

And again higher CPU score with Run 7, but SM2.0 / HDR 3.0 did not drop like it did when SLi is enabled, or gain much while it has 2 more cores then the E8400. So again there is a very strange bottle neck.

So I think atleast with these scores it is true that more cores doesn’t mean better “3dmark score”.

So the big mystery is what exactly are the bottle neck's and just how to correctly decode these “3DMarks scores” which only Futuremark has the answer…

(Run 4-7 images posted below)

 

JUSTIN YOU HAVE Q9550's????????

 

Not for sale......

 

Is that better than popcorn?

 

Yes!!! I already ate all the popcorn...

 

@Justin,

Do you have access to an M1730 so that you could run similar tests on that system under the same initial conditions? Also, if you do multiple runs on the same setup, how much variation is there in the results? Finally, what OS were the test systems running?

 

heh, thanks justin.

as your scores show. the x9000 with single 8800 is scoring less stock than the m1730 single mode and stock
and even the 9550 in single mode is lower than 1 by alot in sm2.0 stock in single mode

 

id rather get a m9750 with 7950 sli for that same price

 

It looks pretty weird. i`ve also been trying to figure out what is the stinkin` formula used..
The closest I got is Sm2+Sm3+1/2CPU ..(there`s always a few houndreds not worth mentioning).
And I will say CPU frequency matters more than cores.At least on 3dmark06.
Maybe Vantage will prove more benchmarking-worthy

 

While they don't add up exactly they are pretty close. When you say there is a CPU jump in run 5 but not in the score that can partly be attributed to the fact that your SM scores dropped by 605 and 513 respectively while the CPU jumped 1434 so the disparity isn't much.

It not taking advantage of multiple cores correctly is probably correct, but it is still showing some advantage. More than it would in games to be sure.

In testing the CPU is not the limiting factor. As can be seen with run1 run2 run3 a jump of .4 ghz had little effect on the scores and almost no effect on SM. A jump to the multi cores shows a significant jump in score and CPU but not a jump in SM. I'm not sure what part of this is not as expected? It's showing that CPU frequency effects the CPU part of the score and not the graphics, am I missing something?

I've been staring at this stuff for entirely to long so it's possible I am.

 

I still say the m1730 had overclocked GPU's.

 

they didn't he ran it stock for me you can see for yourself with 1 card.

 

The 1730 tests provided by me had overclocked GPU's where stated and no overclocking where stated. I have no reason to lie *shrug*. Plus Dexgo's tests are overclocked as well so it would be a pointless argument.

 

yes he had overclocked GPU's I just looked.

but the same clocks exactly as MINE!

http://forum.notebookreview.com/showthread.php?t=234162&page=2

look

he was running in single mode with the same clocks.

I had the higher CPU score but the m1730 still got a higher overall score and higher sm2.0 score etc.

 

Can you link the post?

 

I just posted the link and re-confirmed that he and I had the same clocks and drivers and both were in single mode

 

No the one with his non-sli, no oc gpu no oc cpu.

 

He already provided a link. I saw a 500 points difference in SM2 and about a few more in SM3. Pretty freaking weird ...

 

I dont have access to a M1730. Very little variation, we used a mid score out of 3 runs on each result. (very time consuming )

Each system was running Ultimate 64.

 

I was not calling you a liar, you been cool on these boards.
I just havent found the one for non-sli, no OC GPU/CPU thats all.

 

I think this is a problem with the 9262 only.

Cus i know people who have the X9000 on the 5793 with 8800m, and they are getting 10K in 3dm, same as dexgos score...

 

Hey Hkman, what`s your score with the X7900 ?

 

If your talking about dexgos link half way down..., his GPU's are OC'd

"Clocks are at 600 core and 950 memory. Some people on the 1730 owners thread have pushed further but like I said I don't want to fry my card ."

 

havent tested yet...

RobHague, who has the exact smae config as me, scored 9.5K in Vista

 

which means, that since he got like 9.2-9.5K in vista, he should be getting 10K + in XP, the OS that dex is using.

 

I don't have any benches handy where neither is overclocked at all. I'll run some later and post results.

I do have one here where I ran with no GPU overclocking and CPU at 3.4 and I got 14290 total 6670 SM 2.0 6749 SM 3.0 and 3108 CPU.

I'll run it without any overclocking at all but it shouldn't make a difference since even with overclocking my CPU isn't scoring as high as the 9262. Expect some scores later today.

EDIT: Note the score above was on XP which I no longer have installed. All other scores are Vista 32.

 

If you can with the single card that would be awesome.
Thanks man!