Yeahh finally received new SLI cable today, replaced it with old Aw18 cable. Now SLI working fine as it should be i guess.
![]()
-
Well it was faulty SLI cable.
Got another issue now with 3940XM
i did overclock it to 4.5 on all 4 cores, then why still it throttle back to 1,198 MHz ?:S
disabled = short power limit
max power turbo = 99
flex = 10
I believe i can hold 1st rank for few days!!! Here is Heaven 4.0 benchmark, preset= Extreme
Cloudfire, Mr. Fox and Robbo99999 like this. -
Robbo99999 Notebook Prophet
Good news Peter about the new sli cable sorting out the sli issue! Looks like you've got an awesome system now, just the auto fan control to iron out now (and perhaps your CPU issue is unrelated to your 980Ms).
Peter likes this. -
Thanks bud, yes finally Dual 980m SLI working without any major issue. I will have to set fan profile to manual whenever do benchmark or gaming. I still think that Nvidia driver is not optimized for Alienware SLI system. Firestrike benchmark with single 980m = 8300 and Dual card SLI = 11457.
-
Robbo99999 Notebook Prophet
Yes, that's not much of a score difference with sli enabled, I don't know why that's happening. If I was you I'd have a google around to see other 980M sli results, I'd start by looking on the notebookcheck.net website if they have any reviews up yet (I can't remember). -
Well i did search Firestrike score on 3dmark results. And found 4940MX with Dual 980 (core clock 1173Mhz) SLI @ 14822. I noticed that on benchmark my CPU 3940XM maximum turbo core clock always at 1174Mhz. I think CPU throttles.
-
Peter and the standard set in the BIOS overclocking level 3 44/42/41/40 and nothing more !
Then see if you 11Mark in hwinfo max cpu 4.2ghz !
do not change anything and the only level 3 standard vid 25!
Then you'll know everything
see that and the result will be greater
Peter asked you about a great help:thumbsup:
Could you give me your presence inf.driver latest gtx980M modified for M18xR2 ? -
well i did set bios Level 3 from the first day i got 3940XM, just wanted squeeze it more
Here you go 344.75 mod. inf -
SLI FTW
72C for 3DMark11 is respectable as well for beastly GPUs like this. Congrats on finally getting it to work properly. Here`s to hoping I will follow too soonPeter likes this. -
yea i believe it require little more tweaks. Yes just upgrade it man it bro..you will love it. I heard that you were lookin for 970m but i think 980m is better option and price is not much greater compare then 970m
Cloudfire likes this. -
is well see how high mast Psychics Score almost 11700k:thumbsup:
is very well I have 4 ghz I only 10600k
In that case, I order two gtx980M
Cloudfire likes this. -
GTX 980M is tempting, no doubt about that. Its $1500 vs $900 though.
-
Huh where do you get 2 X 970m @ $900? i think you can get 2x 980M at $1400.
-
RJTech. Seems like they lowered the price for 980M down to $720. Still $1440.Peter likes this.
-
Actually $1398 with 3% cash discount :thumbsup:
-
Rjtech has always been at 720.00 per card.
-
710, 750, 720. Its been fluctuating.
But who cares, still a lot of dough. Gotta pay to be the best. I need to think about it
Have anyone gotten the cards over +135MHz or whatever the limit is btw? -
Peter your physics score is more or less in the ballpartk, my 4900MQ with 4 cores @ 4.2GHz and single core @ 4.4GHz gets about 11200 physics score in FireStrike. You can ignore what 3DMark reports as the clockspeed, because half the time it would report the wrong value in my experience.
However your 980Ms are scoring way lower than they should be. Compare your graphics score to HTWingNut's P377SM-A 980M SLI review sample:
Even with a 4810MQ his sample scored 19126 graphics, compared to 15220 on your machine, so your 980M SLI is running 20% slower than it should be.Robbo99999 likes this. -
He's right. Your 980M SLI is underperforming in that bench.
This is why I don't like 3DMark 2013, it doesn't calculate the score properly when run separately (Fire Strike). If I run all the benchmarks I get around 1.5K higher. It's very annoying. -
Robbo99999 Notebook Prophet
Have you monitored your GPU clock speeds Peter when you were running these tests? Also monitor GPU usage of each GPU during the test. I use GPUz to do that - creates a graph you can look at afterwards to see clocks & temperatures (but you probably already know all about that anyway). That might explain why you're getting a low score.
EDIT: You wanna see a stable 1126Mhz (think that's the max boost clock) on both cards, and you wanna see high GPU usage on both cards throughout the test. Firestrike scales really well with two GPU's with 980M sli results I've seen, so you should be getting close to 100% GPU usage on each card. Have a look at the following link, 95% scaling on 3DMark11 for 980M sli, it's in one of the posts on that page (post #488):
http://forum.techinferno.com/alienw...o-980m-alienware-m17x-r4-m18x-r2-aw17-49.html -
@n=1
Yes Dual 980m SLI is not performin as it shoud be. I did check Clevo system benchmark with 980Ms, they are way higher then mine. I think 980m dont like Alienware yet
@ Zero989
Im goin to run all benchmark now and will post it within next few mins.
@Robbo99999
Yes i did bud. there is no GPU throttle when i set manual fan profile. I'm going to perform all test now again. -
Robbo99999 Notebook Prophet
Cool, check your GPU usage level too though, not just the clocks.
Peter likes this. -
Meaker@Sager Company Representative
The card does normally throttle more with clocks rather than usage but it's certainly low.
-
Alright just ran all primary tests, Results improved finally
Fire Strike: Score 13255, GPU 18735, Physics 10615, Combined 4700
Sky Diver: Score 29593, GPU 53782, Physics 10211, Combined 19395
Cloud Gate: Score 26604, GPU 94531, Physics 7569
Ice Storm: Score 156462, GPU 363563, Physics 52263
Generic VGA video card benchmark result - Intel Core i7-3940XM,Alienware M18xR2
Mr. Fox likes this. -
^ well that's improved, but still not as good as it is in the Clevo, which is a problem since your CPU is a lot better... unless you're on windows 7 and that's screwing with firestrike?
-
See? I told you. 3DMark2013 is trash lol. I stopped using it.
Peter likes this. -
yeah it is. Actually this is what i found on 3DMarks results, NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7-4900MQ,Notebook P375SM highest at stock clock 1038Mhz score 13440. Well it is lightly lower, now it is time to feed system with dual 330W power supply.
Yes sir your right
I do always run only Firestrike, but bug noted now
-
Haha okay. Good luck then, though I don't think you'll be able to overdraw a 330W brick without a modded vBIOS for OCing. I think stock's max is +135 on core =(. Of course I've no idea how much your 3940XM drains, so... meow. I'm going to keep quiet now XD.Peter and Robbo99999 like this.
-
Dual 330W PSU? are you even maxing out a single 330W PSU? If so what is the remaining wattage showing in HWINFO64?
Also from the looks of your Physics score it looks as if you have applied no OC to the CPU.Robbo99999 likes this. -
lol Actually there are some top scores with dual 980Ms. But i noticed that all scores are with Desktop CPUs. When you check graphics score, it almost nearby to my system. Im goin to get RMS digital meter to check how much power my system draws.
-
Robbo99999 Notebook Prophet
Peter, don't bother buying another 330W PSU for your system, I'd be surprised if you need it. Buy a Kill-A-Watt meter instead (cheap) and see how many Watts your power brick is drawing from the wall first.
(Plus, I find it useful to use for system monitoring when I'm gaming, all I need to do is glance at my Kill-A-Watt meter and I know from experience how efficiently it's functioning in terms of CPU & GPU usage - for example I know if I'm gaming at 150W power draw that it's operating at 100% GPU usage along with a good chunk of CPU usage too (games coded well), so if it's instead chugging along with low fps and 130W then I know something not 'right'.) -
Im goin to find out with RMS meter. Remaining wattage showin in HWINFO64? how to check?
OC to level 3 -
yeah i got check first how much system power draws in total. Game Damn i still didnt install Steam yet. I got to test some games now....
-
I uploaded my [email protected] 980m sli@stock benches in TI.
Writing with celluar. sorry too hard to give you the link. -
Yes seen it at TI impressive, finally our system running with Dual 980ms.
I still wonder how can you keep your 3940XM @4.6? i tried bios settings but wont work stable above level 3. -
CPU OC is a whole different story.
I dont use the bios "pre-set" OC levels. The settings are not good. I'd would rather phrase it this way. "it sucks".
And every CPU has different ability, meaning every CPU runs at different voltage even at the same clock (or even at non-OCed stock clock).
That's why you should play with her. Spend alot of time with her to know her better. You should find the "best" voltage for her.
The highest voltage m18x R2 MOBO can supply to CPU is 1.501v. And within this voltage, how high she can be OCed will be determined.
Mine can be OCed up to 4.9 but not stable.
I have rock-stabilized XTU settings of mine for 4.4, 4.5, 4.6 and 4.7.
And I've succeeded to pass Lynx 0.64 avx version 20 times at 4.6. So I am pretty sure my m18x R2 [email protected] is ROCK-Stable.
Intel Core i7 3940XM @ HWBOT
Go visit hwbot.org and see the XTU settings of mine and Johnksss. But they are just "samples" which may not (or will not) fit your CPU.Mr. Fox, bumbo2, D2 Ultima and 1 other person like this. -
Yeah, I agree with this. 3DMark 2013 is kind of a retarded benchmark. It frequently makes mistakes in calculating scores and it's probably the least meaningful Futuremark benchmark tool in existence. It's like a software toy for people that pay no attention to the accuracy of scores. 3DMark11 and Vantage are a lot more meaningful and valuable, and they provide a better understanding of system performance than 3DMark 2013.
The problem is having to use the worthless pile of excrement known as the Windows 8 OS, you can pretty much forget about meaningful benching fun. 3DMark 11 physics score will be about 1,000 to 1,200 points lower than benchmarks run in Windows 7. I think having to use UEFI without Legacy makes 980M SLI a crappy upgrade option because I cannot get Windows 7 to run that way. I may pass on upgrading until there is a vBIOS for Alienware that supports legacy BIOS and Windows 7.
As Brother godfafa said, factory OC BIOS settings suck. You need to use your own settings with an unlocked BIOS or your CPU is going to run like a pile of junk and get way too hot.bumbo2 and godfafa_kr like this. -
^ I bet Mr. Fox is just staring with a watering mouth at Clevo waiting for the X99 successor to the P570WM, going "ANNOUNCE IT! ANNOUNCE IT!"
Mr. Fox, TBoneSan and godfafa_kr like this. -
Yeah Peter, as my " hyung nim" said, you gotta find your own setting for your own 3940xm.
Actually, I've read threads (written by Mr Fox and Johnkss and alot of other OCers) here and those at TI about OC.
I've spent alot of my time researching about OC in laptop especially in m17x R4, m18x R2 and AW17.
You need to find the sweespot for your own girl
-
Thanks for all info. Actually i tried most of setting from Bios, did not from XTU. As i thought Bios OC is more stable but i got to use XTU now. i can go up to 4.5Ghz all 4-cores stable. I will try more tweak later on.
Also i wanted to ask you about R2 fan profile. I noticed that primary fan wont kick in till 72C reached, also GPU throttles at that temp. I dont know why it is happening. when i do use manual fan profile, no throttle at all.
I noticed that CPU temp. wont go beyond 80C at 4.5 @x4, but i just cant OC it beyond that. I got to play with little more
-
My 980m never got near 72 with stock fan table, yet.
I dont like OCing gpu because I had an experience of bricking one of my old 780m. Without OC, 980m never got above 72 while benching all 3dmark11 and 3dmark, not to mention playing games like WoW, diablo and LOL.
I will check and let you know if they throttle on more hardcore gpu bench. -
Hey godfata, how is temps vs GTX 780M during gaming and benchmarking? 980M hotter, cooler, or the same?
-
I dont understand what's wrong with my system, we both got R2 at A10 unlocked bios. Then why my system fan wont speed up till primary GPU temp. reaches to 72C, it's only speed up after that, till then damn quite. it never occurred with dual 7970m.
I also noticed that it only happens when i do benchmark (3Dmark, heaven). Yesterday night i played COD AW for about 10min, both Gpus temp. stayed under 55C (gpu usage 35-50%). Even i dont like to OC GPUs, already replaced 2x7970m before. -
@Peter
Is your how to on first post up to date?
Your are use the A10 unlocked bios Version? -
@Sergeant, Yes sir, procedure is up to date. Yes A10 Unlocked bios.
-
Robbo99999 Notebook Prophet
With the 35-50% GPU usage that you quote it's not going to heat up is it, you said yourself it only throttles at 72degC when on auto fan, and you were below 72 degC during gaming - I'm sure it will throttle if you gamed with a 100% GPU usage though when on auto fan table, you proved that before. Maybe you can find out what's different between yours & godfafa's system & settings by comparing them, that might give you the answer as to why you have the "auto fan table GPU throttling bug" and he doesn't. -
Alright i just maxed out all setting to extra in COD AW, Now both GPUs usage around (68 - 80%). When primary GPU temp. reaches to 63C fan starts spinning and secondary @ 70C. Auto system fan profile holds primary GPU temp. at 65C, wont let it go beyond that. On Secondary, fan is very quite and holds 2nd gpu @71C. Played around 10min. and no throttle at all. Hence i think auto system fan profile works perfect with COD AW, not with benchmark software. I'm using Evga precision x after very long time, it's gone better now
-
Robbo99999 Notebook Prophet
Come on Peter, I think your powers of deduction are better than this, you said it throttles while on autofan at 72degC, yet you only managed to get the GPU to 71 degC when playing your COD game so I don't really know how you can come out with your statement that I put in bold type. You only had a measly 68-80% GPU load in that COD test, what would have happened if that was higher - well your GPU temperature would be higher and it would throttle while on autofan at 72degC. There's nothing special about benchmarking software (apart from Furmark) that is going to make your GPU throttle, it's the "auto fan table GPU throttling bug" that you identified that's the cause. That's the way I see it, I wouldn't expect the throttling bug to just disappear when gaming, just by virtue of it being a 'game'.
EDIT: Unless your "autofan table GPU throttling bug" has magically disappeared. Maybe it's gone and you've found the cause? -
I can't compare them directly now because I've been using liquid pro on 780m but arctic mx4 on 980m.
I had to disassemble about 100 times on this months. You know how painful it is to disassemble m18x R2.
I am too tired to do one more time to re-paste them with liquid pro.
Maybe later.
Aw m18x R2 Dual 980m SLI upgrade!!
Discussion in 'Alienware 18 and M18x' started by Peter, Nov 12, 2014.


