The 9262 vs XPS1730 3Dmark06 score fiasco

right. and he is using vista. if I use VISTA with my card ocd I am only getting 10750 score.

when I get 11500 in xp

I doubt clevo will do anything about this stuff.

I wish we got on engadget like the ALienware.

but people would be like Sager...! Clevo.... Who are they...?

 

Thanks duane123

 

The formula for the total 3DMark06 score is:

3DMark Score = 2.5 x 1.0/(( 1.7/GS + 0.3/CPU Score )/2)

GS= Average of sm 2.0 and sm 3.0 scores

Link:

http://www.extremetech.com/article2/0,1697,1912116,00.asp

 

I just would like to see higher sm2.0 3.0 scores. because they are low. in comparison.

 

Thanks, I already knew that, but I was trying to explain it in a way in which we`d all see how the CPU affects the score.
In the original formula the CPU seems to impact even more. I was just throwing guesses around

 

Guys, I started this thread merely to point out that at stock speed the cards perform similarly and that the NP9262 will come ahead in 3Dmark06 due to its CPU. Overclcoking results are not the issue here. What I would like to see is a comparison between the 2 systems at stock clock speeds. Since there are a few people with NP9262 running SLI and since duane123 has been kind enough to run benchmarks for us, I would like to see benchmarks between a NP9262 and XPS M1730 running at stock speeds, both is SLI and single.

As far as why the Dell gets higher scores with similar overclcoked speeds as the NP9262 is not the goal of this thread.

 

No problem I am in the process of comandeering a desktop at this client site so I can run these benchmarks while I continue to work. Oddly enough I don't think they want to pay me to do it WTH!

 

Awesome so from this we can see that the SM scores are indeed important and the major component of the formula. So SM score comparisons between the 2 same model cards is valid.

More benches coming soon.

 

All right, with all these already shown, is this more of a hardware issue or a vbios issue? Couldn't you flash an xpsm1730 8800 bios to see if its a hardware issue? Or is that not possible?

 

overclocked or not. they were the same drivers and clocks.

that is a main part of the fiasco. and duane said he will provide the results of stock.

 

not even remotely possible.

I cloned the VId-map tables into my 8800m with their voltage table.

and my card severely underclocked itself and Black screen of death/artifacting in windows desktop and freezing..

I personally believe it is a Hardware Limitation with the Cards.

 

Probably not the right thread to post this, but everyone posting here seem to know how to compare 3dmark scores well, so can you tell me please if the score I got on my new laptop that I just received today sounds logical based on the specs?

M570RU
X9000 @ 3.2
8800GTX 600 core, 900 memory

3DMark06 score:11213

Thanks

 

Very Great Score!!!!

On vista or on XP?

 

As you can see, this score proves my point

5793 + x9000/x7900 + 8800m is greater than or equal to D901c plus qc and 8800m. Maybe its a problem with 3dms recognition of quad core, i dont know

 

XP of course, I wish I could get that much in Vista only with a 600 900 overclock! Unfortunately it freezes if I go more than that, did not get a very good pick on the card.

I will try to tweak windows a bit as well. This score was with all processes running (incl. firewall, antivirus and much much more).

 

nope the CPU score it BANG ON! for the QUADS, i compared it to others on the net.

the GPU's in the Clevo or the Chipeset new MOBO... SUX.. somethings Definently Wrong with it.

cuz that's the score I get basically -200 or so points but OC'd 600 1500 950

 

the freezing is a problem with the clevo cards. they need the .14 bios and a volt mod in your 8800m to go past 600 900

 

and that's totally the same as the m1730... x9000 and single 8800m

the score is about right too

I don't think it's something wrong with the cards..

I think it's the Laptop! now that I think about it.

the bus is locked or something is restricted past a certain point... per card anyways.

that's why I hit a wall with overclocking. when others can get higher results.

I haven't opened my laptop for over 3 days.

I am mad at it!

 

Its a problem with the gfx card....

Can you run ntune stability test?

 

You would have to see all the scores to say that. His CPU score will still be significantly lower than the 9262, or at least mine always is vs dexgos. We got basically the same score in single card overclocked (mine was 11581) but his CPU score was 700 points higher. That is what started this whole debate. It looks like if he bumped his memory clock a little bit he could get close to the same numbers.

If you could run it at 600 core/950 memory and post a screen for comparison that would be great. The m570 uses the same exact card yes? That will help us figure out where the difference is.

My non overclocked numbers are still coming I've been caught up in some work.

 

Remember that the P965 does not officially support quad CPUs or 1033 FSB. While it can be forced to work with new CPUs, usually the latency of the chipset (Intel's term is "strap") has to be increased. This is why, when overclocking, there will sometimes be a drop in performance when FSB is increased. My guess is that Clevo had to do something similar in the BIOS.

This could be done by running some memory benchmarks and comparing them with a standard desktop computer with 965 and 800FSB.

Also you have to understand that laptops are not really made for overclocking, they have termal constraints and much more restricted BIOS than desktop motherboards.

 

the cooling in this machine is different than other laptops.

way different.

this could be true, but it seems people with the e6850 have lower scores than the quads so....

 

I can play crysis @ 1680x1050 and get 35fps on high.

for hours upon hours and never downclocks.

my performance is fine.

3dmark scores are the issue.

we all know that 3dmark means jack all for realworld performance

look at what the 8700's got in sli on 3dmark. and I could tool them in every single benchmark with my single 7900gtx oc'd to 650/800 geometric delta clocked. card in all resolutions over 1600xxx

look up my old theads.

my card works great for gaming.

just not 3dmark

 

lol, So why all these arguments just for the scores, if they have no value? Just a waste of time right?

 

exactly

This has been proven before

some people just gotta get mad about something

 

The nerve of some people - my boss is the same way; he gets apopleptic if he sees this website on my work monitor - go figure.

 

Who knows? Maybe now that you geniuses are getting the issue narrowed down, eleron911 could be conscripted to push, pester, nag and harass Clevo/Sager (and I mean that in the best way, eleron911 ) to do something about it or give a very convincing explanation for why it can't be done. Eleron911 did pretty well (with a few others, too) on the issue of the driver for the fans on the 579x, to which Clevo/Sager responded, so perhaps this issue might be push-able as well. At the very least you guys have done most of the really hard work of narrowing down the possibilities.

 

Ok here is my 1730 no overclocking at all on the GPU or the CPU. 174.31 driver vista 32 bit.



The number difference is obviously less staggering because we are dealing with smaller numbers. However the difference in percentage on the SM scores is about 4-5% where with everything overclocked we were looking at 5-6% so it's very similar. Obviously overclocking exacerbates any differences in the hardware so that would explain the extra % or so.

Clearly not a big deal number wise when not overclocking as the 4-5% barely shows. I'm still very much looking forward to seeing some overclocked SLI numbers from 9262 owners.

 

here is mine completely stock

 

I think the XP vs Vista difference is showing some here as well dex. But over all the differences are still a lot harder to see with no overclock to make the problem stick out. Unfortunately that will probably be a reason that the problem will be ignored because they have point at the stock scores and say they are close enough.

 

I think it would be equal if I was using vista. but the quad balances it out... kinda dumb.. but what can you do.

3dmark scores one thing but performance in games is probably on par 100% as 3dmark vs real world is not the same at all.

the frames I mean in games is equal i bet.

we proved this before using 8700s and 7950's also on desktop with 2900xt vs 8800gtx

3dmark uses an unrealistic scoring method of an unrealistic gaming engine of an application that is not a game.

some people don't care about real performance vs 3dmarks

right now I just cared about the marks. but real performance is fine 100%.

 

Well the reason I say the difference is showing between vista and XP is because I was comparing the scores justin got using vista to yours using XP.

True 3dmark doesn't directly equate to real performance but it's hard to get a true comparison using anything else. At the end of the day comparing doesn't mean much anyway. As long as you are happy with your purchase it doesn't really matter what others have.

 

Any one answer this? I was getting around 10400 with spec below using the feb release drivers. No overclocking at all. Then tried some drivers with modded infs from laptop to go, all of which gave me lower scores. Went back to orginal drivers & scored under 10000. Re formated & fresh install of Vista, updated as before SP1 & any extra hot fixes & now the best I get is 10040.

I've lost around 400 points, yet right now my system has no clutter!!

Any ideas?

sorry don't yey have sli (card on way)

 

Did you roll back your drivers, uninstall/reinstall?
Could be some residual files lingering from the two different installs.
Also when running benchmarks try to make a list of all the things running/in the toolbar. You could of had one extra program up that you didn't before.

Also 3dmark scores vary from run to run. Its best to take the average out of 3 or 4 when trying new drivers or tweaking system settings.

 

Have people noticed that justing is using the Q9550? That FSB must be doing something good

Trance

 

Great score! Higher than several 1730's with dual 8800's.

Can you give us a breakdown of the sm2/sm3/cpu scores?

That was run at 1280x1024 right?

 

It may have to do with the fact that you did a fresh install of Vista. Vista has a new I/O prioritization scheme that optimizes itself on-the-fly by, in part, "learning" about the user's usage habits so that I/O prioritization can be adjusted to give more priority to the visible parts of the apps etc that the user typically tends to use; e.g., if you habitually run a particular app every time you start the system up, Vista will eventually start giving more priority to that app (or at least the visible/audible parts of it) at the expense of other, non-visible services and processes, such as hdd i/o, and will prefetch more of that app's data than it would normally prefetch (Win-XP uses a prefetch mechanism, too, but Vista's is more aggressive and is called SuperPrefetch).

Amusingly enough (if you have a sado-masochistic streak, that is ), Vista's prioritization scheme actually reduces the total system performance, as MS itself has noted. Basically, one of the ideas underlying the new prioritization scheme is to give higher priority to the bits that the user can perceive, like visuals and sounds, and to give lower priority to things like writing to the hard-drive, even if that reduces the overall throughput of the system and thus makes the system's performance less than optimal. MS's whitepaper on the topic is: I/O Prioritization in Windows Vista.

So, if your original scores under Vista were taken after you'd already "broken in" that installation of Vista, the OS would have already adjusted itself to your usage habits, so it may have given higher priority to either the benchmarking app itself, or reduced the priority of other background services and processes that would have ordinarily impeded the performance of the benchmark app. Under a fresh re-install, however, that "optimization" would no longer be there, and it's therefore possible that, because Vista hasn't adjusted itself to your usage habits yet, that it hasn't reduced the priority of any services or processes that are slowing down the benchmark.

End result: run the benchmarks again after you've re-established your usage habits with Vista and see if your scores don't improve.

 

This makes it seem close enough that the difference is in the drivers. 100 difference on the GPU score.

Seems like CPU score is weird with 3dmark06.

The overclockability of the dell seems to be pretty awesome. and overclocking the extreme processor seems fairly beneficial.

The fact that +1500 points on the CPU score only giving a +600 on the total score is funny.

 

Like I said the difference persists when me and dex use the same drivers and shows much more when you overclock. And also it makes sense in a benchmark made to simulate 3d game experience that the GPU is weighted more than the CPU. Neither the CPU in the 1730 or the 9262 is getting stressed in any game, even crysis. And I'm pretty sure 1500 gives more than that anyway, if you refer to Justins tables where the CPU score jumps the SM Scores drop by a ton.

 

Err hes overclocked, overclocked the 1730 gets 15200 or so and stock 12700 or so with SLI, and 11500 or so without SLI overclocked.

Like I said if he can bump his clocks to 600 core and 950 memory (pretty much what we have been using in testig) we can get some greate comparisons. It should imrpove that score significantly.

 

dudes, can someone with SLI please give me a copy of thier bioses?

please?

I want to map them..

 

You know, I would chime in , but as I don`t have the 8800M GTX card yet, there`s no real evidence I can get upset upon
Once I`ll get it, dexgo will have probably demolished Eurocom

 

I hope they listen to me/us too regarding the 8700M GT in SLi It is just not working. Next time I try/buy a game I will be afraid to test it with SLi disabled, I might find that it does not work once again apart from giving good score/FPS and jttery game play at 1400x1050 or even lower.

Trance

 

I noticed he's oc'd, and I know he's running a completely different machine than we've been discussing, and I know he isn't running SLI (what have I left out?)... but I'm still curious for my own reasons. (Well, I'll tell you the reasons. 1) I've seen too many scores posted that were run at 1280x768 rather than the expected 1280x1024 so I needed to confirm it, and 2) His system is not unlike mine and his results might influence me to oc rather than add a second 8800)

As to 1730 stock SLI scores... they're all over the lot (much like the 9262 non-SLI scores). I found them ranging from a low of 10600 to a high of 12606 as reported in the 1730 Owner's Lounge threat, and I'm sure, if I had looked further, I would have found an even greater range. But this 20% range doesn't surprise me much, and it is for this reason that I question the significance of comparing single samples from two manufactures. Your GS score is about 2% greater than dexgo's. Considering the fact that you're using Vista, which seems to reduce this score by 2-3% we might say that your GS score is 4-5% better. (This difference adds a similiar percentage to the 3DMark06 score...400-500 points.) But, does this mean the Dell graphics implementation better than Sager's as dexgo maintains? I don't know, of course, but, while I believe dexgo has shown us some serious problems in overclocking the Sager, I believe that, because of the huge range in scores we're seeing now, it's too soon to conclude that the average stock Sager is graphically gimped.

Good news for you, dex! Did you notice in Swiftnc's post that, with one driver, his CPU score went up 10% when he turned on SLI? That's worth about 200 easy points in 3DMark06!

 

I hope they listen to you too, Trance. Every time I read your posts I cringe because I had a Dell 1730 with dual 8700 cards and the experience wasn't pretty. Dell was nice enough to take it back after all attempts to fix it failed. There seems to be hope, though, because a lot of the Dell guys are getting their problems under control it seems.

 

As I've said before random 3dmark scores are kind of pointless because they may be running stock drivers, bloatware from initial install, etc. A lot of the difference will be in the CPU as well, the CPUs in the 1730's range all over the place. Magnus is running a 2.4ghz where when I overclock im a full 1 ghz higher. So obviosuly that effects scores quite a bit.

4-5% is about on par with what we have been seeing difference wise. Maybe upwards of 6% when overclocked. This isn't a huge difference and is about on par with what I have seen in desktop cards. So it may just be the nature of the beast. I think we have exhausted all the testing we can really do.

Both laptops are clearly top notch, unless anyone asks for any specific testing I guess we can probably consider the issue closed untill there is some kind of update if any.

 

Yep!

Two comments...

1) It's unfortunate that we're benching these machines at 1280x rather than 1920x where SLI really shines. Hopefully FutureMark will give us a free version of 3DMark08 that runs the higher resolution.

2) The more information we can gather that "Dell is better than Sager" the bigger the club we can make for dexgo to use to beat Clevo/Sager over the head.

 

you betcha!

 

Did you get any sense from _Dell what was causing the stutter with the 8700s?

 

Not really. I just turned the resolution and quality settings way down in order to get reasonably smooth gameplay.

I had several other serious problems with the machine too (possibly unique to my machine) which couldn't be fixed over the phone, so I simply decided to return it.