980M in a M15x feedback

Just thought I should share my final feedback on the 980M running in a M15x.

Although super fast this card is not suitable for the M15x laptop. I really recommend against anyone with a M15x getting one.

• The card runs too hot. Even with liquid metal thermal compound applied the heatsink and twin heatpipe are eventually overwhelmed by the heat. This despite running at an undervolt at 1V. Bottom line is that the 980M contrary to written specs is NOT a 100W tdp card.
• Uses too much power meaning an undervolt is necessary for stable operation.
• Due to the above the card has to be deliberately crippled in games where the card is taxed (underclocked) and no chance in making any sort of OC.

The general consensus is that the 970M is currently the best card out there for the M15x and this is correct. I wanted to try the 980M and this is what the bare facts are.

It is shameful the way Nvidia manufacture and release higher and higher TDP cards for mobile and demonstrates their inability to deliver credible performance increase without sacrifice.

The card will be sold soon and we should expect no better with the next gen top 1080M card. Even if it is advertised as a 100W card it won't be (980M is around 130W TDP and with the undervolt 120W TDP).

Hope this post helps put things straight for those wondering where the limits lie for the good old trusty M15x and where to save big bucks it has been fun though!

 

Actually it's 125-140w on 1.075v.

 

The power consumption is obscene really. I know what a 100W card should run like, and this is well above that. I think going forward only the 970M is viable since the advertise TDP is like 75W when we all know its pushing close to 85-90W. I can barely keep the 980Ms in the mid 70s load in the M18x, I can't imagine in a M15x. Too hot, too power hungry. Need more efficiency here.

 

What about the bottleneck with the 920XM?

 

@King of Interns Thank you for this very helpful feedback! You took the initiative and the $$$ to try it out. Not many are willing to do this.

 

Indeed, good work KOI. I think we need a list of all tested cards and their caveats in the M15X for future reference.

 

Unless someone make custom heatsink, fully made of copper and 3 heatpipe. I don't think 100+ w is possible on M15x without extreme throttle or need for nerfing voltages.


CPU bottleneck wise, that's the only thing we can't compensate for in the M15x, since the socket is not compatible with other CPU. But at 3+ ghz it should be enough to not bottleneck any GPU considerably.

 

SVL7 posted about creating a custom copper sink for the M15X in the TI forums. Not sure if it was completed though.

https://www.techinferno.com/index.php?/forums/topic/686-custom-gpu-heatsink-design/&page=1

 

He did but it still only featured 2 heatpipes. 980M needs 3 heatpipes. 2 heatpipes no matter how big the sink will eventually get overwhelmed when a gpu consistantly uses over 100W of power.

It isnt a big problem. We just need to stick to 2nd or 3rd tier gpus not top tier.

Cpu bottleneck is not a severe issue at all. 60 fps+ is possible in 90% of games where cpu is a bottleneck. Another reason why 2nd tier next gen card will be a good match up with OCed 920xm.

Sent from my SM-A500FU using Tapatalk

 

You are welcome. Have to put your money where you mouth is

All MXM cards work in the M15x. The only thing that might help is to classify the actual TDP of the cards. IE those that will use 100W plus should be highlighted as red in a list to discourage people getting one. I can think of the following cards:

780M, 880M, 980M, M295X, 580M (plus equivalent quadro cards) are not suitable.

680M and M290X(7970M,8970M) push the limits while 970M seems the right balance of power consumption and performance.

 

I would just add 680M at stock is nowhere near the limit. When adding a few more volts and +192MHz core then you are probably getting close but my 680M ran very happily at 1GHz core during long gaming sessions. Cooling mods do become an essential at that level of overclock though.

 

Adding volts beyond 1.05V in general doesnt bode well with any card in the M15x

Sent from my SM-A500FU using Tapatalk

 

1v is a 25% increase in voltage, NOT an undervolt.

yes, exactly, should have read further down, my bad, haha

actually just have to say, King of Interns doesn;t know what he is talking about.

he is most likely using a SVL7 vbios that is overvolted to 1V+ pushing the card to its limits.
(Breaks more cards then it is worth it IMO, unless your chasing numbers rather then playing games- remember - benching pushes videocards to 99% usage, that's enough by itself to cause damage over time.)
Every so often, I get someone coming to me with a burnt out 980M wonder how it happened, and the vbios usually shows overvolted to a volt. Mind you never happened on the vbios's I used. Because they are NOT overvolted, they are more efficient instead. Usually never rising above 65C when in full load.

 

Wow an unsolicited attack out of nowhere. Gonna report that! Surprised you liked it deadsmiley.

Iam using a 1.00V vbios specially made by Prema for my card (ask him if you want) I have never burnt out any video card and have run many.

The 980M is simply too hot to run in the M15x.

I would love to know where this attack came from...in fact I will also ask Ethrem to confirm that Prema makes 1.00V UNDERVOLT vbios for the 980M.



Sent from my SM-A500FU using Tapatalk

I guess because you sell this card you think I am undermining your sales? The card is simply not suited for the M15x. I am not out to take away your business but to make sure people dont buy the wrong card for this machine.

Having run 5 different cards in this machine over the same number of years and never killed any(sold all to upgrade) I KNOW where the limits lie and WILL inform others. I suggest you get out of this thread thank you.

Sent from my SM-A500FU using Tapatalk

 

What happened here?

You guys should also take into account that nvidia uses dynamic voltage on their GPUs, so 1v or more than 1v on stock can be observed.

 

You don't know what you're talking about.

SVL7 VBIOS undervolts to 1v while over clocking card to 1200MHz boost which makes a lot of them unstable and Prema's VBIOS is based on the typical Clevo ASIC score which equals 1.068v if I remember correctly. The P377SM-A which was refreshed to use this card regularly hits 70s and 80s under load. It's a hot card and as someone who has been through 4 of them personally and had to turn to Prema for help with the temps by getting a custom VBIOS undervolted to 0.987v, I can assure you that you're misinformed. The best card that I got out of the 4 had a stock voltage of 1.018v when boosting and that was a cherry picked high ASIC pair. Even at stock they still went over 80C in Heaven on max fans with IC Diamond in a 20C room with the back propped up 3/4 of an inch off the desk. Stop drinking the Nvidia kool-aid. Oh wait, you sell these cards, of course you'll defend them.

 

As someone said, the 980M and 970M is just a way for nvidia to dump the lowest binned GM204 silicon from harvest.

My AW18 can't cope with 1.075v, regularly hit 86c on fans full blast. Glad that's sold now, even more that the machine is now EOL.

 

Svl7 vbios actually runs at 1.1V load ( according to gpuz) on my card at 1200mhz.

Prema's exactly 1.00V. No what surprised me the most was the ferocity of the attack and accusations. Anybody reading this thread from the top can easily see this is just a user feedback thread for M15x owners only.

M15x only has a dual heatpipe and fairly small sink. It was designed during the era of 75-85w cards not 100w+. Of course it struggles with the beastly 120w 980M (this at 1.00V). Normal voltage puts it at somewhere around 130w!

Such behaviour is against the good nature of this forum that I have been a member of for 8 or 9 years.

Sent from my SM-A500FU using Tapatalk

 

You can't really say that for certain, because even with mod bios there will be a variation in voltage due to ASIC quality or other factors. The GPU operate on an adaptive voltage not static.

 

This is true. However gamble that is asic is not a great situation to be in. As Ethrem mentioned even cherry picked very high asic cards ran at 1.00V. Also high asic means alot of heat! So with the limited cooling we have it doesnt help much.

Sent from my SM-A500FU using Tapatalk

 

Voltage is fixed on both VBIOS mods. Apparently svl7 must have updated his, it was originally 1v and 1200MHz which made it totally unstable for a lot of people.

Prema went the other way and compensated for the average Clevo ASIC quality.

Either way, no stock 980M will run under a volt when boosting. The voltage bins are between 0.85 when idle and minimum I've ever seen is 1.012v under boost on a really high ASIC (80-something). The average 60ish ASIC will run somewhere around 1.07v which is what Prema optimized his mod for.

Either way woodstackz is just plain wrong. It's also an asinine statement to say that benchmarking kills a card because of the high GPU utilization. There are plenty of games that push close to 100% utilization, fire up Bioshock Infinite and max it out and let it sit there with no frame limiter or vsync and watch the utilization. High utilization is because of optimization and a video card that fails because it is being used efficiently is just a junk card, end of story.

 

wonder what vbios does woodstackz use?

 

Umm.. I didn't see any post as an attack. I thought we were helping each other here. Aren't we?

Sent from my overpriced Galaxy S6 Edge +

 

probably misunderstanding, since generally 980m require high voltage + added variable voltage

 

The heat issues you guys are referring too, are with overclocked cards in overclocked situations with overvolted vbioses. Also being in a very old laptop, it's not fair to suggest the cards are BAD for the laptop, 970M also a Maxwell, works great in it, and keeps temps cool and even with little to no effort and there is not as many overvolted vbioses out there confusing everyone, so temps stay low.
Though I will mention I forgot how high the voltage was on the Maxwell cards for moment, I was thinking like the desktop variants which were 0.8V and up ( At least that is what I think mine is at, currently) while the laptop ones the M versions were above 1v. but do infact operate at lower voltage when they throttle or are underclocked - but not easily. I just want to say - that 80C is high for these cards, even in a M15X. They should be around 70-75C in better cooled laptops, they should be 60-70C. Yes the old M15X runs hotter and has some power issues, but which Alienware has no issues, LOL.

Yes, you are right about the 75W/80W heatsink. More favorable for a 970M IMO then a 980M. If ambient temps are in the high 25C+ then it can be an issue with 980M, but otherwise if the ambient temps are lower, say AC is on cooler 21-22C rooms would keep the laptop cooled enough to run the 980M decently, all things considered how old it is.

@King of Interns Hey sorry, no hard feelings. I was wrong plain and simple. So, forgive me ol' chap, I made a silly stupid mistake while I was tired and not thinking clear. Got completely mixed up, following one thing and the another. Sometimes by the end of the day when I've peered over specs on 10 different laptops all day running different configurations, I don't even realise how tired and confused I can be until I see what I posted and facepalm. Hard. Also, didn't mean to come off as a douche, something going through my mind ticked me off or something, I do not recall entirely.

So yeah just wanted to weigh in, and apologise too. I haven't had a M15X in my hand in months, and haven't owned for over a year, and didn't have it long either. That being said, I still forget much about it, from time to time, since I'm focusing heavily on 3-4 other laptops at the moment.

 

Believe me, I didn't over clock my 980Ms except for quick benchmark when I first got them. Sager has my machine right now, a P377SM-A built for the TDP of these cards, and they scored a result of 88C with an auto throttle of roughly 10% down from boost clock with the laptop not propped up and that went down to 81C and no throttle propped up an inch in the back. This is using Unigine Heaven and IC Diamond. These cards *do* run hot, in some cases as hot as the 880Ms they replaced (which I've had two pairs of as well, in the third RMA I fought for 980Ms), it's a problem because Clevo had a high ASIC run for awhile. I think mine are 82 and 77. High ASIC = high leakage = hotter running and would have made a nice overclocking chip for a desktop but shouldn't be put in a laptop using heatsink meant for a 100-120W TDP chip, much less a machine like the M15x which I believe has the same heatsink my M17xR1 had which is to say, it's amazing they got 780M even running nicely.

Thank you for apologizing to @King of Interns as well.

 

Thank you for insulting me Ethrem. Benching is much more then just running a game without a frame limiter. Heavy load and 99%+ on the GPU when overclocked and facing near max TPD does infact kill cards, because we test them and break them all the time.I do not know what specifically does this, on the card 100%, but you will notice the ASIC% drop 2-3% after 12 hour sessions on some cards. I took a card from 81% to 67% ASIC from benching with it all the time. Another card I used and left games running , games like the witcher 3 for days sometimes weeks, and it nearly maxed out the GPU utilization, and yet didn;t lower the ASIC % or damage the card. Both on the same laptop. I personally couldn't get the card under normal stock usage to stay at 100% usage for long without forcing it with 3rd party programs, for days at a time. So I wouldn't say my statement is extreme stupid or foolish or asinine as you put it, but all in context. In this thread they are saying they are easily hitting 80C already. Anyways, wish you a good day Hope this thread is not derailed, was not my intention, I just wish not to create extra fear for M15X owners. I think that was my original issue (though i made some mistakes, I was tired I am sure...anyways already apologised )

I still recommend getting maxwell personally, for the M15X I rarely recommend 980M above a 970M for it, but the 980M that pass through my hands from time to time for the M15X have always been careful made/flashed to work as best as possible with the M15X. It would be a different experience taking a card meant for say, with the usual decent Prema 84.04.22.00.EC vbioses for Clevo's and placing that in the M15X. Maybe a stock 84.04.22.00.10/11 in worst case, should still work well too. I think the M15X has limited power, so you wouldn't be able to OC the CPU as much too, if you get an overvolted vbios on the GPU working on the laptop. I just recommend going all stock or vbioses made for the M15X when upgrading it. Temps, limited heatsink/cooling/limited power and PSU's plus a very aging design amongst other aging driver concerns, all make the laptop on its last breath so to speak IMO. It did have a good long run. Think it handled upgrades better then any Alienware to date did, actually. Even better then the M17X-R4 all things considered. ( at least IMO)

 

Oh we were simultaneously posting, anyways, water under the bridge, I was clearly not in proper frame of mind and tired - it's happened before. I pull 30 hour tech support days sometimes.... then pass out for a day or two. really sorry.

Hey thats very odd, to go through so many 980M RMA's. I got through thousands of 980M. I've had 4 RMA's in closing close to 1900 980M cards from my current vendor. I'm actually serious. No joke. I swear to god. So, something is wrong there. Or some crazy fluke - as statistically improbable as it is, I've been really lucky. Maybe QA could be stepped up somewhere, but AFAIK or suspect there is no way you go through the same volume I do, but having so many personal RMA's is a bit odd. I once had a MASSIVE spout of RMA's with a 780i FTW motherboard with EVGA back in the day. Went through 5 boards. 3 completely new systems. 2 of them being shared components from one setup to the next. That was so odd, I went down and visited this guy , Mike Fierheller in person outside of EVGA (think he now works at Thermal take mind you) to discuss the random issue's. Like it was a big thing. These board JUST came out... but that's how odd it was, so had to say something, because it just wasn't right. it's not like they had a stock of backstock/refurbished boards... these were new. So that situation comes to mind when i hear you have experienced many RMA's



Oh on side note. 880M in M15X = NO NO ! I do NOT recommend that card in sensitive laptops hahah. Disaster waiting to happen. M18X-R1 or like maybe a M17X-R3 sure, they can handle it with triple pipe heatsinks and more power to handle it too.

 

ASIC quality doesn't degrade over time... at least not that I've ever seen between 4 faulty 880Ms and two faulty 980Ms. It's a number set at the fab that is used to determine the voltage the chip requires. It isn't a user modifiable value and it doesn't change.

Basically what you're saying is that these cards are only reliable as long as you artificially limit their performance. You're basically admitting that they will all fail in machines like mine with 120hz screen where not many games maxed out can do 120FPS so they're running full throttle all the time. Since Pascal is basically a Maxwell die shrink that runs as hot or hotter than Maxwell did and they're shipping them with 120hz Gsync displays I guess we should expect massive failures from them as well.

When I buy a GPU, I fully expect to be able to use it within its designed parameters without it failing but by your explanation, it would appear that I should run my cards with no AA (since it's settings like TXAA and especially MSAA that push these cards to their highest utilization and temperature) and cap my framerate at 60 as well, effectively making my 120hz panel a waste of money.

Stop making excuses for the failures. Nvidia put out a garbage product in the 880M and then quickly swept that under the rug with another garbage product in the 980M. You hardly ever hear of a 970M dying and those are the cards everyone gets and overclocks the living daylights out of to get the best bang for their buck. Nvidia needs to stop using high ASIC chips in laptops.

 

880M in anything is asking for trouble

I am well aware my failures were not the norm, especially considering that I very rarely did any heavy gaming and my sessions were relatively short (less than 2 hours). My guess is it was a motherboard issue and Sager just replaced my motherboard due to another issue on this RMA so we will see how my 980Ms hold up this time. I mean I run a lot of obscure games that don't even require a high end GPU. I have this stupid game on the desktop that keeps my 780 Ti at 324MHz even at 4K and I play a lot of older titles like I just started Bioshock 1 with DSR upscaling to 4K last night and while it boosts my 780 Ti to 1150MHz in some areas (water and fire effects), it's mostly running at 836MHz and the temperature stays between 51 and 71C (max the card goes is 82C from what I've seen when running a demanding game which isn't bad for GK110, especially considering EVGA running such high stock clocks on it).

Honestly all these failures are Nvidia's shoddy engineering, not misuse of cards. Unless you're pushing 1.2v to the card 24/7 and running 90s for temps, you really shouldn't be able to kill a card.

 

970M is perfect for the M15x.

980M not so much. I replaced the heatsink and fan on it when I installed the 980M and used liquid metal (which works fantastically). However under full load in a game like crysis 3 the temps just keep rising until the card crashes. This process takes about 30 minutes due to the metal tim but just proves the heat is just too much (more heat created than dispersed)


Also no worries woodz. Tiredness can do that. Water under the bridge.

Sent from my SM-A500FU using Tapatalk

 

I had the same issue with the large copper heatsinks in my P377SM-A- the liquid ultra couldn't keep up with the low rate of heat dissipation and eventually after about a half hour I'd end up at the same peak temperatures that I would get before liquid ultra so I stopped using it because it's too much risk to the GPU if it's really just holding off the inevitable.

 

With a lower gpu thougj the stuff will make a difference as the heatsinks will become effective earlier on meaning lower load. The maximum heat dispersion rate must however remain higher than the heat absorbtion rate.

Bigger sink or higher airflow needed. Probably combination of both!

Sent from my SM-A500FU using Tapatalk

 

Ok so I'm not the only one who finds the 980M runs HOT. FFS, in my Alienware 18, I can barely keep them under 80 degrees with fresh pads, paste, and slight elevation. even mid 70s makes me uncomfortable.....680M-7970M really were good cards..neither one really ever ran above 70.

 

No you're not even close to the only one. Reports all over.

 

We'll see what Polaris/pascal can deliver..hoping for a repeat of 7970M-680M type deal, without the early failures (per 7970M)

 

If these can't do it... yeah.

https://www.google.com/search?q=P37...=chrome-mobile&ie=UTF-8#imgrc=11OEzXjZVhR8UM:

 

Pascal is a fireball, don't expect much. The days of low temps seem long gone in favor of higher performance.

 

It will require beefier heatsinks and a higher CFM fan. In all honestly though like I've said before, NVidia has gotten sloppy with managing TDP for MXM cards (since they now more or less own the spec...). 680M was a strictly 100W card. 780M was up in the 122W, 980M up in the 135-140W....980 at the 200W mark.

Meanwhile, AMD rebranding the 7970M for 2 more iterations, then decide to bring the already notoriously hot Tonga core to the scene...

lord help the mobile scene

 

Time to say goodbye to 980M. 500 dollhairs plus shipping will appear in marketplace shortly

Sent from my SM-A500FU using Tapatalk

 

Good luck with your sale!

 

No, Im not saying what your saying. The cards work fine, and beyond expectation. Older Alienware laptops that are not supported anymore though is entirely different.

I've seen those ASIC numbers change. If it's static, why would that happen? (Could a vbios change actually change the numbers?)

By my explanation I never told you what preferences your nvidia settings should be set at. Don't put words in my mouth. 120Hz screens work, though not always very well on Alienware's. As for whatever you wish to guess on pascal be my guest, I love theory crafting, careful not to be branded a troll when coming up with wild theories if others were not following the details from the beginning and think your spewing "facts" though.

I don't know what excuses for NVidia you think I'm creating. I'm not exactly an NVidia fanboy. But, if you had a motherboard failure (based on your guess, since it's now up for debate if your 980M were the cause of the RMA at all, based on what you wrote.) but then omitted that as the reason why the 980M failed, why did you say the 980M had to be RMA'd without mentioning it was the motherboard that was faulty ? Am I friggin missing something here ? It sounds to me like your scaring people deliberately away from the 980M. Aren't you a moderator, for the Clevo forums ? I feel you are or were scaring people away from Maxwell when there is nothing wrong with them, or that this thread was wrongly blaming the cards for no reason. You later went into explanation on how your desktop card is doing. I share similiar examples, an old DX9 game I run, even with 3 x 980Ti can not get 60FPS on it, and it's from 2004 or something... NWN2. No matter what, the game is just not optimised very well, and no matter of stronger graphics is going to give you better frames then the older 8800 Ultra's did. because an older 8800 Ultra was giving me 70+Fps at 2560x1600 back in the day, without SLI. That is because that card was optimised for that Direct X API's or whatever they call that stuff. I'm no expert there.

90C+ is over the limit for Laptop cards. The sensors only read certain area's they can't read everywhere, and you never know if something else is potentially being damaged, if your overclocking, and not entirely aware of every detail. Stock is a different story, worst case they all throttle or crash anyways before taking seriousI think Desktop card can reach closer to that, not entirely sure, as none of mine go that high before facing other limiting challenges.

Not all people are as capable either when it comes to performance. I bought a bummed out 5820K from someone who said no matter what they did it would not overclock well, and he couldn't go past 4.3Ghz. I gave him 300$ and I have all 6 cores at 4.6Ghz stable, in benches and gaming, in fact the PC has been online stable for 2 months. Low temps and like 1800's in XTU (only about 10% less then you'd expect from a 5960X) but you know, the guy went on and on about how "pro" he was and such. I never admitted any claim to being so crazy proof an overclocker, though I have an 8ghz record and 7Ghz record, on desktop and laptop Haswell CPU's- I purchased his "low quality 5820K" and I'm more then happy with it. I've since been offered double for it. My point though, is clearly the Chip was not to blame. but I'm no expert here, you have a whole forum full of them here though, pro's on almost everything. I've picked up tricks and met people in real life from these forums. Great place to be. Wouldn't be anywhere else.

I'm personally rather cautious in my OCing and such. Never go much further then 50-55% on things. Like my 980Ti's I don't run past 1600Mhz. Don't even know how high they might go. But I'm comfortable with where they are, they're safe and sustainable, and since I purchased them second hand for a bargain, and have no warranty and they're nothing special, don't want to risk going nuts on them. I recommend the same approach on laptop videocards, but if your going to overclock and do aftermarket upgrades, know what your doing and be careful. Don't go into zones your uncomfortable with or can not predict behavior, let the crazy "pro" be the ones to test things and make mistakes figuring things out, and then read up on their success's and failures and learn why. Some of why I commented here, was because I saw some of the (what I personally believe) to be the wrong Idea about maxwell cards.

And yes, I do sell maxwell cards.(I'm not writing in this thread to have anything to do with sales) I sell ALL kinds of computer hardware, almost daily, and have for years, but people just come to me for this stuff, it's not like I have a shop or I'm some expert on this crap, but if I know 1 thing, maxwell was a godsend in terms of mobile graphics. All things considered, really. if we go back a few years, if you recall - AMD 7970M and NVIDIA 580M/680M were in the same playing field, they were great for thier day, made headlines ! But 7970M's died by the dozens... like an epidemic had hit them. 680M would last on average a year and a half maybe 2, not that great. 780M was not much more powerful, but at the very least lasted longer. 880M - lets skip them, they sucked. Rebrands that couldn't even hold up thier base clocks easily - and them the 900M series hits. Now, all of these cards are in the same price range TODAY - right now. But on the bottom end, the 7970M hard to find and still tragically die and are unreliable at stock. Same with the 680M. The 780M getting hard to find new anymore too, still expensive. Within 50$ of a 880M or 980M usually, in new condition (not fake new, i.e pulled from a laptop..i.e USED). But all of those are HALF the performance of a 980M but MORE then half the price. Infact 970M might be cheaper AND more powerful then all of those except the 980M.

That is why I personally think the Maxwell were a great value. They also introduce DX10. At least something to go with the forced windows 10 upgrade that is a positive note.

Also, for people not explaining WHY their AW18 are running so hot, it's because the fans are not working properlyAmongst other thing related to the laptop itself). Take the same cards, compare them to your favorite 680M's your referring to, in a Clevo and things change. The thing is 900M series runs cooler then the previous generations. Broken firmwares and bioses from DELL/Alienware are NOT Nvidia's fault. I'm not promoting NVidia, but saying it as it is. AW18 has issues. Context being 900M series doesn't run well with them no matter what because of lack of support from DELL. That's an ALIENWARE issue, not a MAXWELL issue. I have almost a dozen clevo's I can show you right now, running with Maxwell that all stay at about 65C when playing a game. If I replace a few of those with older cards the temps are higher, i haven't of course used the older cards on all the newer laptop models, but the few I have, have shown the 680M/780M/880M all run hotter then a 980M/970M

 

The fans fun fine, both of them. The heatsinks are beefy and well capable of cooling everything but the 880M and 980M. When you add voltage...forget it. Too hot. It's not an Alienware issue at the slightest, as there are other Clevo users ( @Ethrem ) who clearly are having the same problem.

I really would like to know what magic you are running to get them in the 65 degC range because I have never once seen that in any 980M computer running on system cooling along. Rengsey manages to keep his temps down, but he is using a huge notebook cooling pad (U3 I think...)

Also, consider this. If the M15x is able to keep the 680M and 7970M cool eough, which are known strictly 100W TDP cards without voltage mods, how is it can't keep the 980M cool? I think the answer is clear here...too much 980M is a much higher TDP card, consumes way more power, and the cooling can't dissipate the heat fast enough. Sounds like a Maxwell architectural fault here

 

The 680M i could clock to 1ghz with a slight OV using the exact same cooling config (although with no liquid metal) and temps would stabilise.

The 980M i HAVE to force to standard non boost clock 1040mhz or something if I play a taxing game. Or i need to lower vram speed. Stock clocks at 1V simply too much.

I never ran a 780M or 880M so for me the 980M is worse than a 6990M.

I guess 780M was similar to 6990M that I used to run at stock clocks and undervolted to 1V. Temps levelled at 85C.

880M took things to a crazy new level and the 980M somewhere in between 780M/6990M and 880M.

Quite stupid really. The 680M at 1ghz is more or less on par with a stock 880M core. Sick of nvidia

Sent from my SM-A500FU using Tapatalk

 

My X6 has it at 62C right now.... and it's a bit overclocked.
My M17X-R4 used to have it between 65C-67C

I have dozens of users who report thier temps to me after installing thier cards, I even have a thread on this forum where there is a good hundred people saying and showing thier temps for the M17X-R4 and thier 980M's.
I'm relatively sure the P170HM is going to be roughly the same, I know it doesn;t run too hot in my X8 I just tested one at another users install not long ago, and sold a few of these to people on these forums last year, and at thier house, monitored thier temps. Look I don't know why your suggesting you get different results, could be different vendors. I know there is a SMALL difference, but any time I've created threads or posted differences, or submitted proof I've been warned by moderators to remove it, not do it, and it's been taken down. The reason why is because it wouldn't be fair, it would be like showing evidence why one brand is superior to another when BOTH are here on these forums. So, while you get different results and some of the people you know do to, it's not hard to find the results I'm speaking of, they're here, in very large abundance. Now, I had a sales thread that had hundreds of replies on the NBR market place showing 980M results too, but it was removed and my account suspended for a 4-5 day period during the insane crazy gpu weekend we had a few weeks ago. (where most GPU vendors made 25% of their yearly sales). Look, maybe go browse some of the forum slike Tech Inferno, they did a test of this last year and it was concluded the specific cards and vbios that are on the cards I've personally tested and used were better. This is why, I never recommend prema's or SLV7's vbioses, UNLESS your buying them from someone else. Those guys do great work, they really do. But some other companies have 4-5 man dedicated teams working on bioses and vbioses. Not just 1 guy doing it as a hobby. I do not want to start mentioning brands or get into that, it would start to derail the tread.If it hasn't already.

But I can produce thousands of examples if I wanted. The point is, I get it, your trolling me, haha. I just don't get or understand why. What are you getting out of misleading people with this information. AFAIK the 980M have the same TDP as the 880M/780M/680M etc. I also know for a fact the 900M cards are cooler at stock. If your referring to overvolted vbioses and overclocks, then the fact they get hotter and yet don't crash and are more stable at higher temps is only a testimony to thier performance really. At stock - the Maxwell cards are much cooler. I have seen this and witnessed this first hard, one dozens of laptops. If what you have seen or read reported by the crowds your following is different, then we clearly have different results, which might be attributed to the fact 95% of my results are from the same vendor. Maybe yours is mixed or from some rebrand. Maybe its like those cards you see on ebay - NEW, but then they say it's used by saying it's pulled from a laptop, but then insist its new, really, they only played a few games for the last 6 months on it, rarely ever benchmarked it, you can see they only have a few hundreds benchmarks and firestrike scores submitted by that laptop, and you can see the laptop is in great consideration in the photo's the scratches and dirt are only cosmetic, it's really new , or was new in box, when they got it... derp.

Let me change my tone, okay. Seriously, I am quoting my own personal experiences, and those of the thousands who have purchased these cards from me, and or had me install them for them, or the laptops I've had to assemble, bench and test. I thought - everyone was on same page with this. Honestly. I really don't know why your experiences are so different. From what i've also seen, people have LOVED the Maxwell series. Your implying people hate it, or that you do, in some fashion, dislike it, from what I gather - correct me if I am wrong. I know also, there is no real Fix for the AW18, it's broken because DELL never released a proper bios/fix for it. They attempted too, but it didn't fix everything properly, it added some compatibility or something but didn't fix the fan tables - this is what I understood, from all the people contacting me and reporting thier issues etc.. of course when things go smoothly, I rarely hear any issues lol. I also wonder if it's possible driver variance has something to do with our varying opinions too. Can you quote me or show me where people are having issues with these Maxwell cards? cause it's all news to me. Really. Maybe I don't read those forums, but maybe I can help them.

 

Hey, is it possible the sum of your experience with the MAXWELL mobile cards is from laptops like the M15X or AW18, where DELL most notably did NOT help support this generation of cards ? Because that could explain some of this. I'm not suggesting you do not know what your doing, I think you know how to use the tech you've got. It could be a lack of variety though. If I were to relate, it's like you coming to the M17X-R4 threads and talking shop, where i've rebuilt that laptop a dozen times and know it inside and out as well as you know your M15X - our experiences might not be the same out of the box. This is why I wonder - are not Clevo's and MSI on board with what I am saying and suggesting, that Maxwell is indeed more efficient and runs cooler at stock then the previous generation equivalents.

 

Nobody is trolling you mate, we just want to understand why we are getting such huge temp discrepancy, and how to fix it if possible. What paste/pad/torque combination did you use? Did you lap the heatsink? FWIW I have Dell cards, not Clevo, and the problem is also in my system. No way in hell does a 780M, 880M, 980M have the same TDP as a 680M, it just is impossible given the heat generated by these cards has increased from 680M ->780M -> 880M, but came down from the 880M when the 980M was introduced. However, it still runs hot.

Also please stop calling him SLV7. At least get the poor chap's handle right. It's svl7.

 

The only thing that's possible on my end that makes my GPUs overheat is if BOTH heatsinks in my P377SM-A are bad and that's highly unlikely that they would both be bad, especially since this is the second or third pair of heatsinks Sager has used in the system.

My video card failures *may* have been related to the motherboard but we don't have any way of knowing. My last pair of 980M have been functioning just fine, just too hot. Sager was only able to get them down to 81C running Heaven after replacing all the fans in the system, repasting with IC Diamond, and even then they were throttling 10% until they propped the back up an inch to let more airflow in which resulted in Heaven maxing out at 81C instead of 88C and subsequent thermal throttle.

Cards that failed: 880M - first pair one just died Sager replaced both
880M - second pair wouldn't maintain stock clocks at all then randomly one card disappeared from the system, Sager replaced with 980Ms
980Ms - these worked fine but thermal throttled. Sager opted to replace them on an RMA for a keyboard issue.

 

The 980M is just beyond the thermal cooling abilities of the M15x. If the gpu is running at 90% utilisation like it was for wolfenstein new order then I get temps in the 60sC. However when the card is pushed to 100% the heat becomes relentless. It might not be much over the thermal capacity of heatsink but it doesnt have to be. Eventually given time the temperature just rises until it hits 90 or more and crashes game or machine.

680M was certainly below thermal capacity and had some OC headroom. More than 7970M. Both cards however could be cooled at 100% load though.

Also my 980M is a card of yours. The chap who sold it to me here bought it from you new last year. The card is great quality and hasnt skipped a beat for me. It is a beast however and M15x cant quite tame it at full pwr without serious heatsink modifications.

Sent from my SM-A500FU using Tapatalk

 

To be honest, if it was me, with your experience, I would have sent the card back. But now your saying the card was one of my cards and you purchased it second hand. Sounds like someone pawned thier problem onto you. I've always maintained what to expect from these cards, what your suggesting is out of that. If I had to guess, it doesn't have the M15X specific vbios on it.

And Ethrem, the Maxwell series thermally throttles by design when they get too hot, however, the first few months these cards were circulating (Not from me though, I didn't sell them until May openly) there was vbios issues, that to this day give them a bad rep too. The stock Clevo vbioses were great but only for the specific systems they were created for. Very sensitive cards. Back in the day you could take a 780M or 680M and use it in many different laptops, and the performance would be the same. The Maxwell doesn;t work that way. If you buy it for one laptop, don't just stick it in another, contact someone get it flashed to the best vbios for that laptop, and bingo, that should ease up some of the issues (throttling wise). I had the 980M I was using in my M17X-R4 using a 7lvS's vbios because at the time it was the only one I could get working that didn't throttle the card when gaming. Now theres like 4 vbioses you can use on it, for the M17X-R4. Also the one for the M15X is new, came out last summer. August or something if I recall. But your still on a 80W heatsink, just can't be as ambitious.