Aw m18x R2 Dual 980m SLI upgrade!!

Yes it works. Not sure why it's not working for you.
Im running uefi mode and windows 8.1

 

That's good news, I feel lost without Throttlestop. Yeah I wonder why I'm having trouble..? Brother what version are you using?

 

His GPUs were stock, and he tried with his CPU at stock too, throttling in benchmarks. With power management set to "prefer maximum performance". Games don't trigger the throttle.

I also use UEFI and Windows 8.1 and I'm using Throttlestop 7 for haswell and I've got no lockups at all.

 

I was using version 6.

Was talking about him running single gpu and cpu. Giving him an estimate on what he was capable of pulling from the wall. No way to tell till he uses some form of a kill a watt meter.
This is just to add to what others said about needing a dual psu setup.

 

That's weird, so at 68C it will throttle in every program not just heaven? If it does that, we may need a modified vbios. Honestly my 970M has still yet to pass 52C on load with +135/+500 so I haven't been able to test when the 9XXM throttle, although 68C seems a bit low.

Throttlestop is also working fine for me. Windows 8.1 UEFI on my clevo and Windows 10 UEFI on my m18x. I'm using version 7.

 

ThrottleStop works the same for me with Windows 8 UEFI boot as it does with Windows 7 and legacy boot.

Man, with these reports of throttling, having to use Windows 8 for pure UEFI boot, and other nonsense, I'm glad I chose not to be an early 980M adopter. Waiting and watching what happened with 880M was a nice dodge of the bullet, too. It's unfortunate that something as quotidian as GPU upgrades have turned into what amounts to a high tech form of Russian Roulette.

I think I will get a refill on my Diet Coke and popcorn and watch a little longer before ripping into my wallet like a bull in a china cabinet.

 

Didn't we work out that was the "Auto Fan Table Throttling Bug" where your cards throttle at 72 degC, which happens when you don't put manual fan control - I helped you determine that it was the act of activating manual fan control that stopped the throttling even if your GPU exceeded the 72 degC. Or are you talking about something different? (At least that was the conclusion we both came to from your testing a couple of days ago - your observations sound different now.)

 

Well I tried runnin CPU and Single GPU at maximum performance, it didnt throttle at all. Hence, what could reason? for gpu throttle even when i set it to maximum performance. Yes i already ordered Watt meter.

 

Basically i was monitoring through HWINFO64, i think it was my accuracy error. Now im using msi afterburner, which on screen i can monitor core clock. Even my CPU throttles at stock clock, only when dual gups draws large amount of power.

 

Hey Peter, what thermal pads did you use on the 980M?
Reused the ones you had on 780M?
Is the right one on the picture the one for 980M or is it the left one (where you took off all the black stuff around the die?
http://forum.notebookreview.com/attachments/alienware-18-m18x/117676d1415824624-aw-m18x-r2-dual-980m-sli-upgrade-20141112_172600.jpg

 

I don't think the 980m's have the same throttling issues the 880's did. Well not with my 980m's anyway. I have seen 0 throttling cpu or gpu wise... Pushing them pretty hard at 1440p as well.. But it is a different platform so I have to wonder if it is vbios issues on the AW machine..

 

Ignore that pic, it was just to compare Nvidia vs AMD heatsink. I reused AMD heatsink with 980m, pic is for Left heatsink. Yes i reused 7970m thermal pads on 980m (ignore thermal pads on pic. it just old n stupid)

Actually 980m GPU die is big compare to old Nvidia gpus, you will need to take off black coatin.

 

@Peter - if Brother godfafa_kr is seeing steady clock speeds and you are not, there has to be a reasonable explanation for it. Are both of you using an identical vBIOS? If you are not using the same one, might ask godfafa_kr to send you his.

I think it's a power problem, but not a lack of power problem. If you were exceeding the capacity of a single 330W PSU you would know it instantly. The screen would momentarily dim (sometimes) and the system would abruptly turn off and the internal circuit breaker in the AC adapter will often trip. You have to disconnect it from AC power, then reconnect it to reset the breaker. If you were teetering on the brink of exhausting your power source but not enough to shut down, you would surely be having other manifestations such dimming screen, AlienFX dimming/flickering, STOP (BSOD) errors, games and benchmark programs crashing and similar unstable behavior.

If everything is working fine except for the erratic CPU clocks, I would be looking at having wrong CPU settings (including stock, Level 1, 2 and 3 OC - which have never been ideal on an Alienware) contributing, as well as Windows 8 behavior. ThrottleStop should help a ton. I would even look into doing a clean OS install because Windows 8 may have something goofed up in the power tables. Just because you are using the "High performance" profile in Windows does not mean Windows hasn't screwed that up.

 

I'm not sure what to say... So maybe when I get around to it. I might try to figure it all out.

 

Peter have you disabled bd prochot in unlocked bios? If not, why dont you try.

 

Yeah. Something is very odd with the way his CPU isn't keeping even stock. And moreso with the way his GPUs are working. My CPU's never done that and I've never seen anyone on windows 8 or 8.1 who had a CPU just... fluctuate from 1.2GHz to max clocks on various cores.

 

Sorry for hijacking this thread, but is it any areas I need to cover with thermal pads than the red areas I made? If the gap is too big for my 1mm pads, I can just stack them right?
Also as Peter said, I need to remove the black stuff on the heatsink. Can I remove it all? Why is it there in the first place? Why not have the entire area without the black stuff?

 

@ Peter, have you tried using nvidia inspector? You can try and force all of the P-states to use the 1038/2500 clocks and see if that helps the throttling. It could just be the cards going to a lower p-states.

 

He could just force global power management in NCP to "prefer max performance" and his cards should never downclock (even sitting on the desktop). If they STILL do in benchmarks, then it's a definite driver and/or vBIOS issue.

 

You wanna pad all the VRM's (think that's what they are), the little black squares lining the top of the card, and also those same little black squares on the right hand side of the card. Not sure whether you need to pad the slim black rectangular chip on the top left, but wouldn't hurt to do so. Yep, you can stack the pads. I don't know why they use black tape on heatsinks - either way, it's safe to remove all of the black tape on the core contact area of the heatsink. You also wanna pad the 2 black slim rectangular chips that are directly to the left of the right hand side screw hole. (All the other ones that you outlined in red - you should do those too. (Some people might say you don't need to pad the line of 6 black rectangular chips inbetween the 2 large grey square inductors towards the top of the card, but I padded those on mine, better safe than sorry I guess!)).

Picture is worth a thousand words, and would have been quicker for me to just do this:

 

Yes both system on 10A unlocked, we may need to wait until he gets his another 980m. I've sent you PM. I forced it with throttle stop to stay at 4.5. but it still goin down to 1.2Ghz. I ordered Watt meter today and i will get it by tomorrow.

Thanks lookin forward to it

bd prochot? what it is for?

Actually it never happen while using Win7. Im only using window 8 after 980ms upgrade.

 

I've used Windows 7, Windows 8 upgraded from Windows 7, Windows 8.1 upgraded from 8 which was upgraded from 7, Windows 8 fresh install, Windows 8.1 upgraded from Windows 8 fresh install and Windows 8.1 fresh install.

Under every single instance listed above, neither my CPU nor my GPUs have even come close to throttling in the manner you've experienced. I however, do not have 980Ms, so I cannot check this. It's difficult to ask someone who has a Clevo and is not me to do as much testing as I or Mr. Fox etc would test, and post results in a manner I could be satisfied with. I just think that your CPU performance is very weird, and probably is a result of something else that we're missing, though throttlestop running and being turned on should prevent it. As for the GPUs, yours appears to function exactly like 880M SLI is reported to function. Except that in games it won't downclock, though the heavy downclocking in benchmarks is still present. When godfafa gets back a second working 980M, he should be able to do more testing and let us know if your machine is the exception rather than the rule. I hope it is an exception, because then we can fix that, even if it involves getting replacement cards or changing the vBIOS.

 

Yeah you absolutely want to cover those VRMs on the card edge. Those things can run much hotter than the core.

 

Someone actually posted this as a solution in the thread at Tech|Inferno. Apparently, it fixed this problem with his Alienware 17 with 980M. This could indicate lack of complete driver support also.

This is what we used to have to do with GTX 580M to keep it from throttling. Set P1 to match P0 (both performance states set at maximum 3D clock speeds) and it ran like a banshee. There is actually an old thread covering that very topic. Definitely worth trying... nothing to lose.

M18x: NVIDIA GTX 580M Throttle Fix - No vBIOS Flashing Required (Also nice for overclocking on the M17x R3)

 

Well i just set P1 as P0, now it throttles down directly P8 state while doin Heaven 4.0. Just for testing purpose i set all P1, P5 and P8 to P0 state as maximum stock clock. Guess what system freeze for few seconds and turned off it self. Now isnt it power?

 

Only set P1 to match P0 for core clocks and memory clocks. You also need to set the voltage for P1 the same as P0 if NVIDIA Inspector will allow you to. If you try to set the other P-states for 3D core clocks there is not enough voltage and the freeze/shutdown is expected. This would be the same as trying to overclock without enough GPU voltage. You might want to chat with the guy that did this on his AW17 if it is not working for you.

Also remember to set both GPUs with NVIDIA Inspector. Change the drop-down at the bottom and set the secondary GPU to match the tweaks on the primary GPU. NVIDIA Inspector is an excellent tool, but it does not synchronize changes among all GPUs in an SLI system. You have to do it manually when using NVIDIA Inspector.

 

Thanks man.
So basically just put thermal pads on pretty much everything except the core lol. I didnt know the black thingy above the grey blocks was the power supply. I thought it was the row of small black boxes between the grey boxes.

"Boxes" lol, getting really specific here.
Anyhow, pads on everything. Stack em up if too thin.

 

i've already tried P1 state as P0 both core and memory. But it throttle direly to 324Mhz. Previously, i said that i did also try all state as PO, just for experiment Anyway im doin fresh install.

 

Wait, the AMD heatsink from you Crossfire cards are compatible with 980Ms?

 

Yeahhh yeahh yeahh Fresh install win8.1 NO throttle at all ran both Heaven 4.0 and 3Dmark all primary tests. So some or windows 8.1 or Nvidia driver files were corrupt.


Generic VGA video card benchmark result - Intel Core i7-3940XM,Alienware M18xR2

 

Yes Gwapo, i used AMD both left and right heatsink with 980ms no issue at all. You just need to be careful while usin thermal pads.

 

Hahaha, nice mate. Looks like I don't need heatsink at all.

You need a vbios too,right?

 

Yay! Glad that fixed it. :thumbsup:

I was really having a hard time believing 980M was more of a power hog than 780M because it was supposed to be the opposite.

Edit: who is Gwapo?

 

lol i was thinkin totally opposite, as 980m is 120W card. Anyway already ordered dual PSU parts, it is gona be another adventurous

 

Nope no need to for special vBios, just make sure you get 980m with v12.

Firestrike Graphics score 19000 :thumbsup:

http://www.3dmark.com/3dm/4937453

 

GTX 780M is rated at 122W [ link], but it draws WAY LOTS more than 120W when overclocked and overvolted. Running stock you should be fine with 330W AC adapter. Once we get the vBIOS unlocked it will be interesting to see how far 980M SLI overclocking can go with one AC adapter, not to mention two. The dual 330W AC adapter still holds back 780M SLI if you push it hard enough. Based on the physics score under 12K, it looks like you were not overclocking your CPU too much either.

As far as I know, NVIDIA has not actually disclosed the TDP rating. Several of the sources I read indicate 980M and 970M TDP has not been published by NVIDIA, possibly because NVIDIA is focusing on "efficiency" bull crap with Maxwell. It's nice that it looks like efficiency did not end up being code word for "crippled" this time. All indications are that 980M should draw less power than 780M, thus they are able to use it in some Ultrabooks that are physically incapable of effectively cooling a 780M.

 

Actually as per notebook check both 780m and 980m as 100W, may be they are to early. Actually i wann overclock CPU but cant make it stable at 4.5, tried throttlestop but still bounce back to 1.2. I noticed that on physics test it stayed at 4.5 on all cores during whole test.

 

Run a Wprime 1024 test too..

 

Can you post a screenshot of how you have ThrottleStop setup? If you are having problems, try running ThrottleStop with the Log File option checked and post the log somewhere like Pastebin.com - #1 paste tool since 2002! so I can have a look. Include GPU monitoring in the Options window before you start logging data.

Make sure BD PROCHOT is not checked. This can cause CPU throttling problems on some laptop models while other similar laptops may not have any issues.

 

Thanks Unclewebb, I did unchecked BD PROCHOT. Again ran firestrike it worked for few seconds clock did throttle but this time it went down to 4.1 from 4.5. Unfortunately after few seconds firestrike stopped and got unexpected error. please check screenshot of settings. I think log file it not accurate cosz pc restart unexpected error, like clock_watchdog?

here is log file http://pastebin.com/P9Wan91c

 

You don't actually have ThrottleStop turned on so besides monitoring your CPU, it is not doing anything to help you out.

Click on the Turn On button. Also make sure you are using the Windows High Performance profile and that the Minimum and Maximum processor states are set to 100%.

http://i.imgur.com/HfHozrU.png

 

As I said, BD prochot can be turned off in bios too.

 

Peter - Have you done any CPU stress testing while using the 45 multiplier? Some Prime95 or even some wPrime so you can fully load your CPU to see if it can run the 45 multi reliably when fully loaded.

 

Just for the record (especially since unclewebb) is here. Throttlestop is working fine with W8 UEFI. I am using Trottlestop 6 now. With the freezing before I was using Throttlestop 700 b3.

 

Interesting, that's the version(7.00 b3) I'm currently using fine on 8.1 UEFI and 10 UEFI.

 

And as I already said, I use Throttlestop 7 fine on Win 8.1. UEFI is on too. I don't get why yours isn't working.

 

Single card test.

 

Do you have any specific reason for using 344.11?

 

None. It was the last driver I had installed so I just modified it and ran it.

 

in RJtech, it asks for which clevo laptop I have, so what should I select if i have alienware 18, do i need the screws too?