*OFFICIAL* Alienware M18x Benchmark Thread - Part 2

Spoiler

okay ill give it a try later today. Was at 85 amps and low watts ill try that. So i completed that cooling setup last night ill post a pic when i have time. Its an ac built into a laptop desk with the cool air ported through dryer vent house aluminum up to wire passthrough hole in top of desk and sealed my notebook cooler in a removable way that it sits sealed over top this cool air port. Measured from over notebook cooler air holes was at lowest point as low as 20f or -5C i was pretty stoked... Now have to watch condensation under a long term bench. Any advice on slowing or stopping condesnation John before realizing the condensation forming it almost effected unit. I let it warm back up and turned AC off. Its just fine just need to have a plan for the long benches.... Perhaps make them as quick as possible and then let setup warm back up before starting again.

Okay 70-75 flex 150 amp ill let ya know

 

You shouldn't catch condensation as long as the laptop is on. The coldest part will be the metal on the bottom.

 

Spoiler

it did perhaps its waybi have it... Room is warmer than laptop -5c in a 20c like room the cool air is being enclosed and ported under a sealed up notebook cooler. And notebook has foam between the notebook and cooler so all is sealed air from 40 f to 20f depending when condenser was on.
There is a setting called dry instead of cool or fan on the ac should i use the dry setting instead of cool i think it doesnt cool though then. Would a light cloth between notebook and cool air help internal condensation. I say its a prob running full because i had it on and suddenly text got all shaky on the screen the whole thing like something shorting. I shutoff removed battery and allowed all to return normal. So is issue that room temp is so much higher than the laptop the top even felt "freezing" like 30ish at lost little over ha ha i froze it i read temps i quoted with a regular thermometer direct in air flow out of notbook no laptop on top of cooler.

 

You only need to remove the bottom cover and feel if it's wet wet. not ice cold, but actually wet. Check the memory modules to see if they are dry. If i let mine sit, it will drop the cpu temp on core 3 to it's lowest point of 3C. And the bottom will be extremely cold...

As for this shakey text...are you running stock or over clocked?

 

Have you seen this, guys? Looks like the IB has nothing to offer

 

Spoiler

sorry under neath th bottom was wet considerably and i beive the mobo area was lightly touched with condensation too the text thing was only that one time on the bios boot screen tells me not overclock... Never happened again once i was careful of condensation. 5c was reached for me but could get that low had it had longer. The entire laptop got to about same temp cold air comes out of keyboard i have the whole setup sealed so good.. It all removable and portable but i set notebook cooler down then laptop and good to go

 

Hummm, then you might have to seriously watch it then. I don't have that problem

They could be flubbing the scores. Since running 4 cores at 3600 mhz should see about 27.95/13.4k in that chess benchmark. But since it's an es, it might be down clocked some.

The wprime score should be in the 7 second range not 13
3dmark06 is the only score that is right.
3dmark11 should be 8600 for cpu test, but that's hard to say depending on the gpu status.
Edit: re ran and it should be 8600 or so.

So at this point...they aren't fully matured "stock" benchmarks.

 

Yes but the point is, I don't see any "arch" advantage here. Just a 200mhz bump at base clocks and 100 on max turbo. If that's all, - not impressive at all.

 

You would need to check the DT version to find out that...
And just like over there, it isn't doing all that great, so the same will more than likely apply here as well.

 

I'm not expecting Ivy Bridge to be all that impressive from what I have seen and read. We might be surprised, but I think it's more hype and hoopla than anything else. I am expecting to see more improvements with the IGP part of the proc, which really holds minimal interest for gaming enthusiasts.

 

Exactly. The CPU part is less than a 2920 ->2960 bump. I know, it's early to say and the tested samples are all ES right now but still, there's virtually no arch changes - same cache, similar clocks, etc. The only thing that might spark some interest here is a (possibly) lower load temps.

 

Spoiler

i agree that only real benefit is temp and die size one can fit more transistors in same chip size... So despite same architecture maybe you could see mobile 6 cores who knows... The shrink is a nice amount. See 980x and 2900xm series use 32nm one is previous gen yet being same size architecture helped the 2900xm series gain. When we saw i7 first gen @ 45nm vs 32nm first gen they said heck lets throw two extra cores in lol. Maybe since no architecture change but die size shrink maybe we finally see hexa or octa core mobile. Or octa core desktop chips. Honestly if you have a sandy bridge ivy bridge should be skipped and get best of both worlds in the next gen with a Architecture improvement.

 

I don't even think it's gonna run much cooler but let's be optimistic

 

Spoiler

yeah who needs the cooler part as much as it being able to complete more work in the same time.... so i am for the gen after IvyBridge myself.

I measured -5c / 20F with no laptop on this cooler i set up when AC condenser was on.. coming out of notebook cooler surface bit of a breeze coming out of keyboard with all this setup

wrote about making here
What do you use as cooler on your notebook? - Page 2

Spoiler

not first of its kind but it was fun

 

Hey guys! The NBR F@h team has made the top 850 teams worldwide. Our next big milestone/goal is the top 500 teams. Our team is growing each day, so if you want more information, check out the link in my sig. The M18x would be a killer. Especially those with 2920/2960 CPUs and 6990s/580s. Come out and join the effort! We're almost at 10 Million points as a team!

 

^^ Nice number of posts you got there, lol.

I will join the frenzy when the heatsink project is complete. Hopefully, can add some points to the team even without the extra cooling.

 

Ivy Bridge does not represent an architecture overhaul at all - only a die shrink....clock for clock it is identical to what we already have. The advantages will be in temps/power draw and ultimately overclocking headroom because of this. I do not know why everyone expected it to be the biggest thing ever.....

Just like Conroe to Wolfdale in the desktop space....they gained SSE4 other than that exact same core logic and architecture

 

It looks cool. Makes me almost want to put one under my desk as well.

 

Lol thanks. +1

Can't wait to have you! definitely put down some nice numbers with that system.

 

Thanks, 580m in finally got a bit to setup. Need to work on vbios... Never flashed Nvidia. Yeah you could use your portable if it wasnt so tall John

 

Another traitor I see

 

Got my Q1N8 (D2 stepping like Fox's) in last night and its a beauty Holding 4700mhz on all 4 cores. Here's a new Wprime run:

 

Nice score, M8! You need to slash another 17sec to beat a stock 3960X

 

Whos M8? Joker that's great to see it over 4.5 you run vantage at 4.7 yet? so is this the one or more to go through?

the grass is greener it actually is

 

I switched to AMD when felt that the grass was lacking a red hue, haha

 

Nice run, bro. +1


Below is a comparison of custom overclocking against standard BIOS Level 3 2920XM overclock with stock GTX 580M GPU clocks and voltage for anyone interested in knowing approximately what their performance should look like without any significant tweaks with the same CPU and GPU combo...

3DMark11: Result: P6318 vs P8439 | Vantage: Result: P24600 vs P32676

The only real tweak implemented with the "stock" settings is P0/P1 being equal for the GPU to prevent throttling.

For our newcomers, the throttling fix can be found in the opening post on this thread.

 

Thanks Mr.Fox i need flash that vbios

 

Best bet (in my opinion) is using stock core and memory clocks with .92v and using NVIDIA Inspector for the "Fox Fix" to OC only during benching. If you stay around 735/1600 or less, only stock .82v is necessary. (More than adequate for gaming, but not enough for killer bench scores.)

See attached... instructions included. There are two ROM files with P0 and P1 at equal stock speeds and voltage. One is stock .87v and the other .92v.

STANDARD WARNING

Third party BIOS Modifications, VBIOS Modifications and other system level tweaks (including fan control) can potentially cause damage to your system. Be aware of this before blindly flashing/installing.

If you brick your system or cause damage, chances are very high that your warranty will be useless. Know what you are doing, you are responsible for your actions.

-​

 

AWESOME! Bro John^^

 

new run
3dmark2001
ImageShack® - Online Photo and Video Hosting
3dmark06
ImageShack® - Online Photo and Video Hosting
3dmark vantage
ImageShack® - Online Photo and Video Hosting
aquamark3
Imageshack - aquamark3u.png
super pi 1m
ImageShack® - Online Photo and Video Hosting
wprime 32m 5.396
Imageshack - wprime1m.png
wprime 1024m
ImageShack® - Online Photo and Video Hosting
pifast
http://imageshack.us/photo/my-images/207/1565.png/
cpuz 8 threads
http://imageshack.us/f/806/4893.png/

 

Spoiler

thats what id like to do hav .92v 3d clocks and the rest stock. Then NV inspector like your guide shows for the benching and maybe for gaming a seperate profile to load probably with the clocks you just suggested. I alrady downloaded the stuff from Widezu's thread mentioned in your thread. Is it safe to use the vbios .92v widezu provided and instead of in windows do the usb boot drive flash? I flashed 5870m vbios that way and was most comfortable with using dos commands... 6970m i used atiwinflash through RBE editor inside windows which i was less comfortable with. The above nvflash thing for in windows i probably wont do. So back up my vbios and how do i issues commands in dos to flash second sli card... I think i only saw Widezus example for m17x R3?
John been tinkering under 52x, 51x,50 for validation purpose only... 150amp is usually needed to have the single core be able to reach that speed like you said. I am trying to get past 5175mhz but may be my max as its freezing often at second id probably see 5180mhz ive been messing with flex from 40-99 and just trouble finding what is necessary... It seems sometimes i just freeze pushing certain stuff but not always from voltage.... If i cant get farther later ill call it good there and start super pi 1m and 32m. Wprime is obviously 8thread... But i forgot what that 3rd cpu bench used in threads the one you got 13secs.

Edit
i boot at 20-25watt msr lock disabled. 150amp. No cores states, no HT, 1 active core, i have tried 99.6mhz x 52 no go but 99.5 (cpuz value) x 52 validates boots. 51x to equal same didnt do well. 50x 103.5 (cpuz value) worked so i have a 5175 validation, ill post if i cant get higher. The succesful one going over 5170 was the 50x 103.5 but 103.6 fails... Perhaps its been feeling like motherboard not as good on baseclock as last.... I have motherboard replacement coming since i cant get any combination to allow dimm d to work... Seems its broken because my first M18x supported 16gb 1600mhz xmp2 kingston hyperx 9-9-9-27 this one dell ram or this type gives 4 beeps if one is in dimm 4/d so in a week or two ill be getting a new motherboard.

terry amazing job! Those are nice scores i was impressed by the 106+ baseclock

 

I only flash the cards in a Windows admin command prompt. Some of widezu69's guide is for systems with Optimus. If you follow the instructions in the zip files you should have no issues. nvflash is designed to be used within Windows and I am not aware of any M18x owner that has experienced an issue with flashing using the instructions. The command for the second card probably would be a "1" place of "0" in DOS as it is in Windows.

The beauty of using NVIDIA Inspector is the ability to make changes to P0 and P1 without having to flash the vBIOS. Unfortunately, it has no effect on voltage. I have not tried to use it yet, but what I have heard about Svet's vBIOS Tuner sounds interesting and worth exploring. NBR member mobiousblack is using it and mentioned it in this thread. Unfortunately, it does require flashing the vBIOS. Not that doing so is inherently dangerous, but it would be a lot nicer to have the same type of control over everything that flushes and reverts to stock on reboot like you can with NVIDIA Inspector.

 

Spoiler

So an M18x 580m sli is not an optimus system? Okay ill use yours above. I just want to have permanent .92v 3d clocks and if flashing in windows is fine for these okay... In a lot of other cases its not recommended over flashing outside of windows. Thanks for the guides... Just use nv inspector after that to overclock. Amd i used Saphire trixx and flashed vbios for gaming. This case use nvinspector for gaming clocks to and vbios just for voltage. I did same on 5870m's. Thanks Mr Fox

 

Flashing in windows is easy now. Plus you can still flash the cards even if you were to some how mess up. And since we run batch files for this. Flashing is super simplified.
nvflash.exe --index=0 -5 -6 580m.rom
nvflash.exe --index=1 -5 -6 580m.rom
save as a batch file
580.bat = master
&
5801.bat = slave

You can make these for stock voltage or stock video bios flashing as well.

 

Running my m18x modded A05 bios 2960xm @ 4.2ghz 44x multi 4 cores, 85w short 56sec 99w long 865pri 0-flex GPUs - p1,p0 matched 0.87v. When running vantage CPU tests the temps on core3 are getting 100c and computer is shutting off.
Currently OEM paste and back is raised and fans on full in HWmonitor. Am I using wrong CPU settings? Or need to do a repaste? First time I have been getting shutoffs due to CPU temps.

 

You're welcome, Brother RV. And, no, the M18x does not have Optimus. (And, I'm glad it doesn't based on what I have read about Optimus, LOL.)

Yes, works beautifully.

Brother Dan - disable short duration in BIOS. My system overheats and shuts down like that with it enabled. Look at my settings in this benchmark. See the description field. You may need to back down a little bit on multipliers unless you find a cold air source, but short duration being disabled may help your temps a lot and I cannot see any performance benefit to having it enabled. If you still have the stock Dell TIM, I would avoid re-pasting. It does a good job of heat transfer and there's not going to be much gained on temps. The factory TIM does not degrade over time as rapidly as aftermarket paste. (Same applies to factory TIM on GPU.)

 

Spoiler

There is a small benefit in short limit. Disable it for now for the benching but decide later if your using it. Not using it keeps same limit on cpu over short period time and long. Above Johnkss talks of playing Skyrim at 5ghz. Although this can be done via only long limit. Using less seconds that stock 21 of a 2920xm would be smarter... Allowing a quick peak but drop you set by the short limit time into reasonable ranges for longterm loads so you dont have shutdowns in other apps either. Could be your distance between two values is a lot. 80-99 is a jump... How bout higher on the long limit than 80 watts and 97-90 watts may be more appropriate in that kind of setup. The watts wont matter if your flex supports the speeds running at any given second. And the multis need to be used to keep speed where it wont add too much heat and cause throttle or shutdown. If you are on unlocked bios and have not flashed your own modded bios most throttle temp you can set is only 90c stock shutdown temp is 100c seems you may have throttled to if thats the case. If your a risky person you can set 110c and 120c instead of 100c... I find raising it allows atleast a peak at 100c on one core without shutdown if more than that it tends to cause shutdown anyway and as John said once these arent true values the 100c stock setting directly on die might only be 85c you should be running vantage at 4.7ghz+ how high can you validate danxbix?

0flex under loads higher than any of these benches can handle 300amp 300watt 0flex meaning no drop in mhz under any load 44/44/43/43 ive tested under extreme temps and should work good. Perhaps your cpu wasnt pasted good from factory. Please tell us the largest difference between any of the 4 cores under load if its pushing over 5c like as high as 10c you need to repaste. Got 5180mhz last night and crash before validation... So ill push for that again call quits on cpuz.

Mr.Fox in my experience i agree about the dell paste 3 factory pastings 1st decent but 10c between hottest core, 2nd was 2cores 10c apart under load... This current stock pasting is by far better than last two... Maybe 4-5c max difference under load and other 3 cores real close. Just monitor under wprime 1024m if core temps are dramatically different from another (remember this is 100% full even load on all cores) repaste if not leave it and look to get some cool air. Got an old ac units even window type you could make a AC powered desktop for your laptop i posted about above who needs LN1 i got idle temp down to 0c on two cores and 8 and 9 c on other two with the ac i found i can carefully play with AC temp sensor and force it to continue cooling down to 32f-20f at the lowest ever since i used a sealed setup freezing air flows through whole system not just fans so entire thing feels like ice time for me to move on to the benching.
dan i did paste ICD24 two different times. First time temps were better than first two dell stock paste. Second time was in between the two bad paste jobs from Dell. And the latest stock paste job from Dell has been best even compared to ICD24 i have feeling its because its hard to get exact even paste after reassembling heatsink and to not get air pockets or other issues like veining of the paste.... If its not broken dont fix it is a motto and a good one.

 

Well I depot'd my laptop and Dell sent it back. Sound and Video work now, great right? Only problem is they were missing 10 screws from the bottom of the laptop! (including the 4 for the screen hinges) and now my keyboard does not have blue LEDs in the mid section.

I have shared my dismay with their service and recommended they refund me so I can order a NEW (not refurbished exchange) machine. I told them if they are worried about loss if business give me a $3500 Dell Giftcard haha

Will keep you guys posted, hoping to get refunded so I can order a new beast with 2960xm and GTX 580m SLI (ya yall got me a little jealous!)

 

^^ Wait till R2, will save you extra hassle of killing another beast jking.

All of a sudden everyone got upgraded, AW support must be cursing this community, hehe.

 

I really was not looking to game the system....had the broken audio so I called and they send a motherboard and new speakers (they are limited in their troubleshooting and computer knowledge....). New motherboard fixes audio and breaks video. I get a 3rd motherboard sent, fixes video breaks audio. I depot, everything fixed but comes back with broken keyboard and missing an unacceptable 10 screws. They brought this upon themselves.....

Side note: Aikimox, 2960xm and CF 6970m.....that does not match up haha

 

^^ I agree, when they start fixing it almost always ends up in a refund/replacement, hehe. They just can't get it right.

 

Spoiler

Thats what happened to me. 2920xm, 6970m crossfire, dual 750gb, 4gb 1600mhz ram to 2960xm, 580m sli, dual 240gb samsung pm830 and 1333mhz 8gb ram... i have my own ram Scook R2 could be amazing but not as good if the underground doesnt unlock its bios then its hard to say but R1 may then be greater than R2 depends... 675m sli in R2 and 7970m crossfire probably will see a 7990m type and a 680m type

That was after many repair and a depot failure etc... Even now they still have to replace my motherboard on this one.

sorry to make you Jelly Scook.

 

BTW, Just tried disabling the short term and got even higher temps

 

Like i said short limit has its uses. What amps or primary plane and what was long limit watts when disabled short limit? Well and of course what mhz and what program?

 

4.5 on all cores, Pri Plane 880, Long 80, Flex 0, Wprime155, running 32M test reached 88C on one core even with maxed fans and elevated back

Room temp is 78F

 

You guys are seriously funny!
These are laptops. Most desktops aren't even doing these speeds with better than stock heat sinks. You guys need to let that go already. Be happy

As to the 5 ghz. That's on my "real" laptop "cooler". That would not be happening on a desktop with max fans anything. haha

Brother Sam:
Thought you we're a happy camper with your laptop, last time we talked....I guess not huh?

 

So, in other words, there's no combination of BIOS settings that would allow me to have 4.5Ghz 4/8 24/7 without overheating on 2-piped HS?

 

Spoiler

thats sounds right since 78F is a warm room temp. Most times im at less than 70F. but basically you are saying with short limit on at X amount of watts same test and situation yields lower temps than using 80 watts long limit? I think truly heat is derived from volts/mhz and load since amps and watts aren't heat related but what causes mhz to drop or be maintained. i am fallible so i can be wrong. When heat is higher same mhz requires more watts so we find we have to add watts to maintain mhz. I used to run my hottest benches without any real cooling just room temp and fans max and John was always running similar benches cooler... for a little it made it seem i needed more watts to do same but in fact the temp was causing more watts to be used and if it passed the watt limit the mhz dropped. I cant say for sure why 80 watt long is giving you higher temps. Can you try another? Compare short limit enabled and disabled but do a test on wprime 1024 m at 8thread then there is less variables and full even load thats very comparable (100% load vs 100% load) you dont need to do the test at 4.5ghz. Just run at 4.0ghz so your sure it doesnt peak over 90c and see which has higher max temp... Were you comparing 80watt long with 80 watt short compared to just 80watt long or are you using different watts?

 

Ok, let me try 4.0 and see what happens...


Edit: Something weird is going on here. Even @ 3.8ghz and 72W Long PWR PriPlane 0, Flex 0, still seeing the same temps. I might need to check the heatsink again.

 

Spoiler

for the most part. Unless the app isnt a heavy load if heatsink thermal paste are fully optimal you can get that in most apps but anything over wprime no... Unless you use a cooling rig like me and John. Thats why short limit is cool you set the time limit yourself and tell it a nice watt limit and have it drop within reason in the amount of time you found wont reach crazy temps in even the worst of apps. But remember the time limit goes back to short limit in only a few seconds if load on the cpu drops for a few seconds... Then your back on the high watt limit... So dont set time limit for too many seconds. Intel default is actually 21secs i say lower if using a rather higher number when compared to your long limit... if set too high or to too many secs it can cause havoc in games but if set with in reason it actually does not cause issues... Its going to high on those that cause instability since cpu tries to run faster more often than one intended.

John your funny too lol i just meant what you spoke of earlier in thread... I know you meant with proper cooling and you meant as i mentioned under a certain Lower load or even less than 8 thread setting a lower watt allowed you or anyone to run high mhz in low load low thread apps but drops properly... How a cpu should under 8thread load and heavy loads... You mentioned gaming at 50/50/50/50 on Skyrim... And refered to it running only 1-2 cores like that since its the nature of the skyrim beast. Temps therefore kept in check and if you had left same settings and ran wprime you probably wouldnt have overheated and only ran at a much lower mhz atleast thats if you were not running with the watts through roof then the cpu wouldnt have dropped and then caused crashing/thermal limit to be reached. I know you keep it all within realist settings so im guessing thats about what you did. As i had played BF3 and Skyrim in a similar manner on my first 2920xm but 50/50/50/49 obviously as mentioned i made it to drop that 8thread speed if under heavier load so 4.9ghz under the lightest 8t loads and steadily lower as load increased. not talking outrageous settings