*OFFICIAL* Alienware 17 and M17x Benchmark Thread - Part 4

Lol, Way to many kids will be around. All my nieces and nephews under 10 years. I think thats when accidental damage would come in handy Though they'd find a way to kill my 6990s and eveything not under warranty Plus my wife may divorce me...hmmm...all my money/time to myself....hmmm... I need to repack!

 

LoooooooooooooL

So true. For the wife and kids.

Last time myson grab the mouse and just sit in front of the black screen typing like dady, so fun ! Until I notice he had too much pencil on his hands...

Sorry, I know this is not related to our serious topic, but for now I can't do better with my laptop. As soon as I can, I'll grab an extrem cpu to challenge all of you guys.

Merry X-mas to all

 

Happy Holidays you guys!

Merry Christmas Morgor!

 

Well ill be folding on my 540m until I can get a new heatsink. My 540m is topping out at 69*C. Hopefully a heatsink solves it. Don't know how to explain it having heat issues all of a sudden. I tried another 920xm and that had the same behavior as my 940xm. At least least its not the chip. But in the mean time my system is barebones and folding for a few hours. When I get back home ill put it back together and try the modded desktop firepro drivers.

I also have a volt modded 6970 to benchas a 6970 and as a firepro

 

Hi I recently purchaced a M17xr3 with 2730qm with 6990m. I thought something wasn't right from the start as gaming performance was realy bad. I then ran 3dmark Vantage to compare with this thread. My scores are P6827
GPU-5826 CPU-18k. I done all the basic checks. Plugged in to PS set power settings to performance. Checked graphics switching is set to ati GPU. Installed latest drivers(from dell). I know that the Discrete card is running because when i set it to integrated graphics performance drops even more. Games run OK but if i turn on AA on to even 2x it is unplayable.

Is there anything else I can check?

 

I purchased a new m17xr3 at the beginning of December and ordered the GTX560M instead of the GTX580M like I wanted. I called Dell to resolve my issue and they wanted $1300 for an upgrade or $285 restocking fee. After some haggle I received a $300 discount to keep my laptop.

I've already ordered a HD6990M from a NBR member but in the mean time I want to see what I can get out of my GTX560M.

My current OC using Nvidia Inspector is 940 core/1880 shaded/1725 memory at 68°. Anything more and I'll get artifacts or crashes.

Is there any way to increase the voltage from 1.0v to perhaps 1.05v?

Edit: My highest 3dMark Vantage score is 12895. I want to get into the 13k or 14k range.

 

you might have to grab a new processor to get that high, those are some already good scores.

 

I've got a 2670qm at stock clocks, I could OC the processor and get the scores. The GPU comes in at 10405 thats what I want to get the most out of.

 

If theres any R2 owners with a 940xm I could use a bit of advice on overclocking it.

Just got a new 940xm to replace the faulty one i sent back, It works fine this time but it seems to be a poor overclocker.

It doesnt want to overclock at all above default x25 multi without a bump up in voltage, It just ends up bsod'ing or locking up at some point.

If i apply an extra .75mv in the bios i can run it at x26 but anything above that makes it lock up or bsod again. Is this normal because I saw some people on this forum reaching x27+ stable without a voltage change.

Ive been experimenting with the TDP/TDC settings in throttlestop trying to get x25 on all cores under load but if i run prime95 it seems to need more than 100/90 to even get x23 on all cores let alone x25, However the throttlestop bench seems to run it at x25 across the cores which seems odd.

My temps are fine, I have pasted with IC7 and temps reach 90c max at 100/90

Is it just a case of having a "bad" overclocking chip or do you think the above is pretty much average?

 

Here you go, Mick, this should help you a bit:
http://forum.notebookreview.com/alienware-m17x/476426-m17xr2-overclocking-guide.html

Also feel free to browse this thread and the previous parts of the Benchmark thread as they are filled with info.

To save you time you can search the forum this way:
type this in your Google search bar

site:forum.notebookreview.com alienware overclock 940xm

 

There's no point to going over 25x because you can't turbo that high without heat issues in real everyday use. Needing more voltage at that point is normal. Its heat you gotta watch. If its benching you want to do then you'll need to raise the voltage. Nothing you can do to get around that. Every chip is different. If anyone is pushing a full 27 turbo on all four cores then they raised the voltage.

 

I cant even push x24 across all cores when running prime95 - 8 thread FFTs without putting the TDP above 100/90,

I dare not go any higher as temps start going above the 90s

 

Thats normal. The money maker is getting turbo boost on all cores. Even 25 at all cores will be to hot and probably needs at least the +75mV to stay stable while holding all cores at 25x. I've done a full wPrime 1024 run at over 4GHz on all four cores with a 920XM, but that was a lot of work. But it stayed maxed out at 28x on all cores the entire run . Max temp during the run? 44°C on one core. All others were much lower One core maxed out at 6°C


= i7 920xm (ES)

wPrime 32M @ 4.002MHz: 6.275 sec

wPrime 1024M @ 4002.3 MHz: 192.72 sec

 

What kind of TDP did you have to use to get them x25 multis across the board?

Your a lot braver than I would be doing that kind of cooling. is it LN or dry ice?

 

Used dry ice, TDP was set to about 125w. Across the board you'd probably have to go to about 105w-110w or so to hit 25 across all cores. That would be my guess, but I havn't benched in a few months. Also depends on your FSB speed. Less at 133mhz vs. 140mhz.

 

What I find odd is i can run TDP 95/80 and use the throttlestop 1024 bench or wprime and get a sustained x25 across all the cores with a 85c top temp but when i run prime95 with same TDP the core speed drops to 3.06 and fluctuates at 22 - 23x multi and temps top out at 90c

I guess prime 95 is a lot more demanding than the other benchmarks and requires more TDP to keep the multis up

Has anyone ever killed the R2 motherboard or processor by overclocking it too far?

 

For overclocking, I don't think anyone has killed a mobo. At least not yet. Especially since you can't raise the voltage much. I'm going to try and bench the 920xm or 940xm with a voltmod on the CPU to 1.5v. If I can find some time.

Prime95 may use a differant calculation method that is more demanding than wPrime which could be why it drops the multi. I've never used Prime95.

I think I did kill a 920XM by benching it at high TDP to much. Now that chip will boot, but it will BSOD. I think it deteriorated after the abuse I put it through. That was an OEM 920XM. I have an ES 920XM and an ES 940XM as well. Also benched a 540M at over 4GHz, benched a 720M and a 520M as well.

 

Has anyone did this 1.5V mod yet?

 

Svl7. Check his 920xm wprime 32 score on hwbot. Just a matter jumping pins in the cpu socket. I did ot for my 540m and though i didn't gain much, I gained a little. Mostly it was pll voltage holding me back.

 

Ah that ever elusive 27K. Can't wait for 7970

 

what are your temps like? Run a Furmark burn-in test and post your fps and temps. I think your 6990 is running too hot

You freaking dry ice potted your alienware? bro, that's insane.

 

Dude that ain't dry ice...that's liquid nitrogen...

 

that makes it even more insane.

 

That's dry ice, in a pot with some ethanol or similar I guess
LN2 would require some serious mobo insulation to prevent damages from condensing water.

 

Oh right. My bad.

 

Yea Used Dice, still should have insulated better but I couldn't be bothered

 

Even with using dice, you can only go so far. The key to extreme cold is so you can raise the volts, which in this case...isnt a whole lot

 

John, Here is the method I will use to try and get to 1.5v I did it on the 540M but I think the clock gen voltage was holding me back

Here is Svl7's guide

Here is a pic of when I was using the 540M Though it will be similiar to this. The additional two wires are so I can monitor voltage with a multimeter. I'll have to jump differant pins for the 940XM. I'll have to check with Svl7 and look at the tables to see what the max voltage I can get to is. Will also need to break out the dice pot. I know Intels stated max voltage is 1.45v so hopeflly I can get close to that. Don't remember what the bios +150mV gets me to so I'll have to refresh my memory and start from scratch. If I ever manage too get time to do it.

 

Now that's an interesting mod. How many extra mulipliers does it help one to get when running extreme?

 

He was able to bench at 133mhz x 30. 24K CPU score in Vantage. And that was stock CPU +200mV from the mod. It's in an M15x so it doesn't have the same voltage options that the R2 has.

Here is some of Svl7's benches

 

Ah yeah, now I remember. Back when I had the Asus.

 

Yup, probably could have used the mod on the G73. Any idea if this would work on the 2960? or can you already adjust voltage enough on that? You can already hit 1.5v right? If so this route of volt modding isn't going to be of any use if you can already max out the safety limit on the cpu voltage. This 1st gen core-i are pretty robust given the proper parameters. Stock is only the tip of the iceburg on thier potential.

 

Hard to say...These mobiles SB extreme's can run with the desktop SB version in benching.

And the max voltage is 1.5V, but with more voltage it could go higher....Maybe. If you have a maxed multiplier...then no amount of voltage is going to make it go higher. The biggest complaint about SB, but it also turned allot of non over clockers in to over clockers.

 

Ahh, that makes sense. So long for the days of real OCing?

 

During the second overvolting session I went up to 32x stable at 133MHz, never did a vantage with it though, only wprime and some of the other cpu prime tests.

 

Nice! Glad you stepped in to add some accurate input. What was your top wPrime 32M time at 32X?

EDIT: got it, 6.056 sec @4.154MHz

Spoiler

 

Yeah, that's it, hmmm... should try to bring it below 6 seconds sometime... but it just gets unstable when raising the bclk. Dead stable with 133MHz.
Maybe I'll try some stuff with the RAM when I find time.

 

Yea, maybe some better RAM. I had my 540M up to 179MHz on the FSB for the 540M, though maybe it's easier as a dual core than a quad? Though I have bumped my 920XM high on the FSB. Maybe just when the multis are that high it gets unstable.

Spoiler

 

Cool... I have pretty decent RAM, 1333 CL7, maybe the timings are simply too tight for overclocking. Will see what I can do in case I find some time.

About the SB CPUs, if you can up it in the BIOS to 1.5V... wow! Pretty crazy. Whether you can go higher, I don't know, you'd need to check the VR, but if I remember correctly the SB chips use VRs which go higher than 1.5V, I think up to 2.0V if I'm not mistaken. But you'd need to check this on your mobo.

 

I believe John had discovered with desktop 1366 chips (exact same core logic) that actually dropping the BCLK down to 100 and using the multi to get higher clocks yielded him higher performance. This is also (coincidentally?) the same thing that intel ended up doing with SB.

 

Hmm and I've been juggling both. Welcome back scook!

 

Depends on the chip really, i had a desktop core i7 920 (1366 socket) that liked a slightly lower multi but higher fsb.

I think a lot of it has to do with the motherboard too.

 

I need to try and break the 5 second mark. im at 5.020 now.

It's always best to start with a low memory clock to allow for higher room to over clock. Then come back and start getting fancy later...

Also, these are still mobile chips bound by the motherboard restrictions on voltage. To much will cut the laptop off. Even if you were allowed to go that high.
hard coded in the bios. The part we dont see. It happens with the 580M when pulling past the 100W (over flow of 10W to 110W pure max - speculations on that of course) per pice slot. That's what stops my over clocking. Not temps or even the .092V but the pcie voltage max.

This is true...

You would have to go far lower than 133 to find out. I had mine down to about 110 fsb. And you should have seen how them test use to fly.

This is true.

 

Hey guys, I recently received a backplate from an M18x:


This was to test on my 580m. The picture and logic is self explanatory. Anyways, I put it one, replaced the other one. Fit nicely. Booted and there were tones of artifacts when posting. Windows also refused to boot. I thought this may have been an error on my part and I destroyed another GPU but I swapped back my original and it was fine. I swapped it 3 times with the same results. I insulated everything making sure that the metal never contacted anything else. Nothing. A tube of IC Diamond almost gone.

Anyways. Take it from me: the L-Shaped backplate is a no no. No idea why.

 

Do you have a picture of your other card and bracket?

And it could be your missing a pad in the right place. I think some others have had this issue as well, but not sure what they did to correct it.

 

My original backplate is this one:
New Alienware Dell m17x r3 GPU Heatsink Complete J9H7X | eBay

In the picture and the bit covers the vram chips along the long side of the card.

 

Did you put pads on the both sides of the bracket?

 

No I might try that

Edit: Nope didn't even post this time 7 beeps which is system failure. Swapped back and working fine. Strange isn't it?

 

You may have messed up the memory chips...speculation of course...
Can you take a picture of the card?

 

*sigh* ok I'll do that later. Just put my system back together and it's working fine. Lemme play some BF3 before I do anything else. It's been a long day.