*OFFICIAL* Alienware M18x Benchmark Thread - Part 2

Well Brother Sam, what do you think? Your input is always welcomed.
It would also mean all scores need to be validated on 3dmark.com.

No problem!
Don't forget to run the other test....

I'll get these posted.

 

This one? Did I do the screen shot correctly? In your screen shot I saw 1M and 32M in separate screen shots. Is it OK to have both in the same screen shot? Does the first test need to be closed to get good results before starting a new instance of SuperPI? (Haven't tested both ways yet.)

 

Sure. Looks good. Just don't need the gpuz screen, because these test are only cpu.

 

New Vantage score also Steve... gonna need to push a bit harder to take back #5 spot on this one, bro.

Vantage: NVIDIA GeForce GTX 580M SLI benchmark result - Intel Core i7-2760QM Processor,Alienware M18xR1 score: P26538 3DMarks

Spoiler

​

Already had it open from previous bench and just never closed it.

 

wPrime: 32M - 36.75 / 1024M - 1150.444

 

Nice run Bro! - I'm gonna have to get ta' tweaking lol

 

Mr. Fox - on the wPrime try setting it to 8 threads now

John - what is the HWBot stance on it? And with ORB it was ambiguous to me, are they saying tessellation has to be enabled for the result to be valid? Me personally....well you know I like big numbers

At the same time though.....unlike the PhysX thing which is proprietary to Nvidia, tessellation IS what is being tested....so turning it off does come across a little like a cheat in my mind

 

I forgot I had the BIOS set to 2 cores and HT turned off for the GPU-Z validation. Will run it again today. Thanks for noticing.

I agree with the "cheat" thoughts. If AMD can run with tess off to get higher scores, then NVIDIA should be able to run with PhysX enabled on principle.

 

All this benching almost makes me wish I had an 18x, but even then I don't thing id have enough time.

 

wPrime: 32M 8.07 sec / 1024M 254.968 sec

Well, you should want one. I'm sure your M17x R2 would enjoy the company. I understand that it's hard being an only child. A little sibling rivalry can be healthy. The little M11x cousin probably is not enough company.

 

Hi all,

I am new to this forum and am also an owner of a new M18X

I have 2 questions.

1 ) How do I get around the Dell video drive issue and use the video drivers straight from the Nvidia website. Dell seem to be way behind in their release of upto date drivers.

2 ) I have run a couple of benchmarks and here are the results.

3D Mark Vantage = P16417
3D Mark 11 = P3707

Both benchmarks say the system is slower than others with similar specs and that the video drivers are not approved.

What can I do to improve these results. Is it likely to be a driver issue or a hardware problem?

Thanks



Alienware M18X, Intel i7-2860QM CPU, Nvidia GeForce GTX 560m SLI, 16GB DDR3 1600MHz, 2 x 500GB Raid 0 SSD Hybrid HDD

 

Looking better.

That's the funny part. The highest gpu score for amd is 8100 for AMD. And the gtx is 12536. Ill have to ask over there to see what they are doing about it. But if you check the first page...You'll see with just me dropping the tessellation by like 40 percent. i get 9100 points for gpu.

 

Quick peeps Get on it lol!
Joker here is one for you!

Intel i7 mobile CPU i7 2920xm ES 2.4Ghz | eBay

Cheers.

 

mpearson - welcome to NBR Forums. We're glad to have you, and especially glad that you're part of the Alienware Community. Congratulations on purchasing the M18x. It's an awesome beast and the finest gaming laptop you can buy. As I am sure you've discovered, the build quality is superior.

Those scores may be just a tad low for 560M SLI, but it's probably a simple setting that is not right.

Before updating any drivers, let's tackle the fixing the low scores. Drivers are not going to make a gigantic difference. Dell's recommended driver actually works pretty well.

It appears you are may not be connected to AC power. If you are not connected to AC power, do so. You cannot game or bench well on battery. Also change your power profile to High performance rather than Balanced or Power saver.

Unless there is something odd about the 560M SLI, all you will need to do is uninstall the Dell driver and download/install the NVIDIA Verde driver. The latest release for the 560M is 285.62 and you can download that here. It is generally recommended to do a "clean driver installation" and if you're not familiar with that process we can provide some tips. Let us know how much help you need and I'm sure you will find ample assistance from our community. The best thread for your technical questions relating to the NVIDIA GPU is the http://forum.notebookreview.com/ali...s-performance-problems-discussion-thread.html. If you have not already read through all the posts on that thread, it would be a great place to start.

I customarily share helpful links when greeting newer members. Please take a few minutes to click through them. I hope that you and other readers find the links useful in jump-starting participation in the Community. Again, welcome to NBR.

USEFUL THREADS FOR FORUM MEMBERS

All members reading this post are encouraged to become familiar with the rules for our forum. The information on how to search the forum is especially useful, as it should help members find information quickly, avoid creating new threads to ask questions that have already been answered in another thread, as well locating appropriate threads to participate in an established discussion.

Forum Rules - *PLEASE READ BEFORE POSTING* | Alienware General Information & HOW TO: Search the forum - *PLEASE READ BEFORE POSTING*

Reputation Points/Power and User Titles | Forum Guide and "How-To" | How To Properly Insert an Image Into Your Post | NBR Marketplace

 

A quick run of the old skool benches!

Cheers.

 

Merry Xmas fellow benchers and benchette's!

 

And you too John ...

 

bah, I am selling mine for 300.

 

Cheers everyone and thank you for all the good advice

 

Mr Fox Thank you for your warm welcome and advice.

I have installed the latest video drivers and the scores have improved.

3D Mark 11 = 4196
3D Vantage = 16647

Is it really safe to overclock the GPU's. I have read on some sites that overclocking laptop components is crazy.

If it is safe how do I do it?

Thanks

 

Ok, so when are we going to see some numbers?

 

Ok Guys and Gals With an extreme cpu. I ran these test to show that all of you should be able to break 30k! Now im not accepting no for an answer! We should have far more people breaking 30k! I am also doing this to prove that you don't need 30k cpu to do it either.

Spoiler

4.5 ghz

4.4 ghz

4.3 ghz

4.2 ghz

4.1 ghz

 

Thanks for posting the configs John.

Why PhysX off?

Also what GPU voltage are you running. I can't get anywhere close to those speeds without overvolting.

 

Can you tell us what your overclocking settings are for the CPU? I have my multipliers set to 45 and I set my GPU to the same as you thanks to that vBIOS ;-) but I still only manage 26000 on vantage... I don't know what is bringing me down. My CPU will not let me overclock the Bus speed at all really. Bringing it to 101 makes it freeze in the BIOS so I leave it at it's stock 100

 

Freq : 3633.62 MHz (103.82 * 35)


------------------------------------
NVIDIA GTX 580M SLI:
3DMark11: P7831 | Vantage: P27343 | 3DMark06: 23762 | 3DMark05: 28434

Spoiler

3DMark11: P7831

Spoiler

​

Vantage: P27343

Spoiler

​

3DMark06: 23762

Spoiler

​

3DMark05: 28434

Spoiler

​

------------------------------------
AMD HD 6990M CrossFire:
3DMark11: P8092 | Vantage: P25730

Spoiler

3DMark11: P8092

Spoiler

Vantage: P25730

Spoiler

Cross-Reference: Cumulative Results with 2920XM CPU | GTX 580M SLI
(Post-upgrade benchmark results with Extreme CPU for comparison)

 

There is always as risk in overclocking, whether laptop or desktop. The difference with laptops is the components are more expensive to replace and harder to come by. You cannot buy MXM modules from Best Buy, Microcenter, Fry's, etc.

Look at the link in my signature for the "throttle fix" and you'll find all the info you need in the opening post on that thread. If your M18x has GTX 580M SLI, those scores are still too low and I would encourage you to figure out why before doing anything else. Those scores are similar to a single-GPU setup rather than dual-GPU.

Have you made sure that SLI is enabled in NVIDIA Control Panel? If it is not, do that first.

 

Humm, I was a bit hasty....I see your 580m's now.
Basically everything I asked killaz

Did you flash the unlocked bios?
Did up set the primary plane to at least 900?
Did you set the vid flex to 0 or 5?
If you do not know what im talking about...it would mean that you didn't read up on this my friend....

Ill get you updated in a few.

Updated!

And the final ruling on 3dmark11 is.....It's allowed as is, but it will need to be ran on the new version of the software. We follow the Hwbot ruling on this. And since it's a software driver based tweak. It will be Allowed.

 

And here are the oldies...

3DMark03: 104543


3DMark01: 52436

 

Spoiler

First page updated.

 

Cool. So wuith the GPU overvolted, were you running in a freezer to get those results John? Seriously awesome!

 

You can run that right off the desk. No freezer needed.

 

Nice! Well once I'm done comparing the AMD and Nvidia systems at the same settings, I'll see how far I can push this Nvidia sucker

 

What are your cpu settings at now, stock?

 

Yeah mate. At the moment I'm still testing my two systems head to head at stock settings running stock BIOS on overclock level 1.

Furthest I've pushed them is overclock level 3 in the stock BIOS with all cores set to 40x. However this doesn't seem to actually work as CPUID still reads 37x under full load with those settings. SO I'll need to man up and flash the unlocked BIOS if I'm gonna push any harder there.

Furthest I've pushed the GPUs is 725/1450 GPU/Shader and 1600MHz RAM at stock voltage.

I'm still getting much lower benchmark results on the NVIDIA system than I could achieve with the AMD system, but in game frame rates on the NVIDIA system completely stock are higher than the AMD overclocked as far as I can push it on stock voltage with no lockups.

It's very strange.

The other weird thing I've noticed is that my wPrime scores on both systems are very different, even with the same exact BIOS settings.

The nVidia system gets 7.5s for 32m and the AMD system gets 10.5s running the same test. I can't figure that one out as it has nothing to do with the graphics cards. lol.

 

Open throttle stop and make sure they are really running the same multiplier and watts. chances are...on is running lower watts or multipliers.

 

Sorry... I totally missed that part of your post. 560M and 580M look enough alike that it just blew right by me.

You can look here to see what the ballpark should be: NVIDIA GeForce GTX 560M SLI - Notebookcheck.net Tech

Your results (GTX 560M SLI):

3D Mark Vantage = P16417
3D Mark 11 = P3707​

Notebookcheck (GTX 560M SLI):

3DMark Vantage - P Result no PhysX 1280x1024 - min: 16294 avg: 16294 (53%) max: 16294 Points
3DMark 11 - Performance 1280x720 - min: 3735 avg: 3912 (44%) max: 4089 points
​

I don't always regard Notebookcheck as the ultimate authority, but having data to compare with is always useful. So, on second blush it looks to me like your results are right in line with what they achieved and it seems you have nothing at all to worry about. So, from here on out you can focus on getting more from your 560's through tweaking things.

Have you applied the newer Dell 560m vBIOS firmware already?

nVidia GeForce GTX 560M
Release Date: 10/10/2011
Version: A0270.26.36.00.09
Importance: Urgent
Download Type: BIOS
File Format: Hard-Drive
File Size: 8MB​

The link in my signature for the 580M Throttling Fix has instructions that may be applicable to improving performance of the GTX 560M. You can try using NVIDIA Inspector to speed things up a bit. It would be totally outstanding to see you posting your results here along with the AMD 6990M CrossFire and GTX 560M SLI.

 

Here are my scores from before my system went down. I've got probably 2 weeks before I can bench again. John can you pm me if there are any issues with the entries?

Spoiler

wprime 32M 5.095sec


Wprime 1024M 167.016


Super Pi 1M 7.363Sec


Super Pi 32M 7min 12.096 sec


3dmark vantage P27168


3dmark11 P8051


CPUZ 5.0906 Ghz
CPU-Z Validator 3.1



alternative super pi 32m 7min 28.366 sec
part 1


part2

 

Brother residualvoltage. I'll get those up in a few minutes.

 

Brother John much appreciated cant wait to get back my system fed ex is picking it up anytime this day. Your 53x is so awesome hell yeah brother John f-ing awesome. So sweet. I was thinking about my methods before when testing and have a feeling i can get over 5090 validated... don't think 5.269 ghz but hey you never know lol. I figured out i had amps and watts to high to boot and was running warm... question is will i get 52x or 53x to boot.

Edit:
BTW I like the avatar photoshop job?

I noticed your highest validation you were on IGP in your opinion would you say the system was more stable for validation purposes only being on IGP, the other thing is you use 1 ram stick and for some you use lower ram speeds... that's also always helpful I think.

 

I hijacked that avatar from the net some where..

It's hard to say. My last chip was a 4800 mhz chip. This chip is a 5200 mhz chip.

I have had amps at 1023 and watts at 600. no difference. It's just how sandybridge is. When you reach the wall, there is no more.

 

true i notice same thing about amps and watts eventually makes no difference like the voltage i have a feeling at either 89 or 90 flex anything above is null. Yeah could be chip limits. For points under the max effective watts and amps i noticed interesting thing.... same watts set and one has higher amps... you will see small differences in that scenario... like a few runs i ran 300amp and 99-110 watts for certain benches and it maintained much higher speeds than at 1/3 those amps and same watts. vice versa the same but not as much seems raising watts and not as much the amps is less effective. Now the third way one might configure is use same amps as watts when benching... my 2 week waits going to be agony... is this wall of fame M18x gods ... end by a certain date? or does it go until the momentum has been lost? I'd love to get a chance to best my own scores.

 

This wall will stay, so no worries. plenty of time to better your scores.

 

so 1.3711 or could have been 1.3761 is max i see @ 4.4ghz @305amps @97watts @100baseclock IBT run @ 0 flex... so max 99 may be close to 1.68V... maybe 1.63~ is the actual max at around 90~flex... but could be between 73 and that though... 73 is max i had noticed cured voltage bsod...

 

Maybe...but I don't think that's how it works. The max voltage is 1.5V. And this flex thing is more for the qpi vtt voltage, but some how raises the actual votlage to it's max point. Even pushing 70 flex and 1024A it still does not go above 130W total draw.

 

okay I see. I think the draw can be higher but at that point the bench or app we would be referencing would be causing thermal shutdown... your right in your earlier statement that's in agreement with a statement of mine about 4933mhz thread being actual max... at wprime load.. only thing that will run higher is intel processor identification utility... its 8thread full usage but its load itself is most likely closer to 1/3 the load wprime 32m would give @ 8thread. i have many runs in that app at 50x 8thread @ 100 baseclock ... but any higher it dropped...like all the others. This limitation i believe is due to the pathways being designed for a much lower load... i noticed too if in 2core 2 thread mode... and benching that each core can go to about 50-60 watts each so 100-120 watt max in dual core... so individual cores must not be able to take over 50-60 each... that seems to support the 205watt peak max in that screenshot for an 8thread 4.7ghz load in IBT for a second or so..... and that wouldnt mean i could go way higher in other apps it just suggests at the max load of IBT.... the most one could see is 4.7~ 8thread and not for any constructive purpose... but it does tend to say to stay below that... so if each core at most can see 50-60 watts max it would support the theory. But in actual obervations it seems more like this 1 core highest 2core same minus 10% watts 3 core again 2 core minus 10% watts... and 4 core same giving a pattern of total power draw per core ever lower as you add threads.... dont quote the watts or any numbers from this but there is a seeming limit of total watts through each core or multiple cores... btw in your opinion John what is most accurate on total watts used? Hwinfo, real Temp, Throttle Stop, hw monitor. either way even if this isn't exactly accurate i'd like to find a way to monitor it that would be considered accurate because to me I am trying to learn all aspects to better understand its inner workings. Strange mine has seen over 130W at those settings but I cant be totally sure since you didn't say a watts for reference. this is via hwinfo and throttle stop. Hmm I just need something credible to monitor with. what app are you using for this?

John thanks for posting my scores. I was surprised to be at where i am with so many scores still to acquire. I did notice my super pi 1m is listed as my 32m run... for sake of others being like whats with that...i figured id let you know.

 

Well. I look at it like this. If my chip hit 150 Watts and all i got was a lowely score. I would not be spreading the news. . That's just me. I take it how it comes. SB chips aren't like the past revisions. I have had a chip under -60C and it still didn't budge. -180C and still no go. I wouldn't get to technical about all this if i were you. Not really that serious. Basically takes all of 10 minutes to find out if the chip is a good over clocker or not. Then you fine tune from there. I only use throttle stop because i watch the watts as they happen. and take the average (not a one time peak) to derive a number. To many what if's in the end. And what happens more so than most. Some one comes in and beats a technical explanation with "all i did was set this and go" A new world record. And at the end of the day...It was just a better binned chip....

Spoiler

 

When you put a load on your chip over 4 cores its still the same in the end? Cpuz says... But TS says the facts. What does it say at those settings at 99-100% load 8thread? If you can maintain perhaps the 2960xm > 2920XM No?

 

What exactly are you getting at?

 

LOL.....this sums it up pretty much.