NP9870-S (P870DM-G) - Nvidia GTX 980 - Overheating

I've been having an ongoing issue with the GTX 980 in this thing overheating. Soon as it hits 70c it's basically done and black screens. I've tested a wide array of things and am at my wits end.

I've reached out to XoticPC regarding this (whom I purchased it through) and they've been less than helpful except for only recently due to my public display of frustration over their customer support.

I'm hoping perhaps some of the professionals here have some experience with this or something more to test? I've tried underclocking and that works for awhile as long as my game settings are low enough to never push it to maximum load, which generates too much heat. Primarily low settings with VSYNC is the only way I can manage to keep it from overheating so quickly.

Below is my XoticPC topic with everything I've tried. There's also HWINFO dumps in my topic with my most recently reply containing 2 temperature dumps showing it crashing at 70c. Even with fn + 1 maxing the fans out it still overheats.

http://xoticpcforums.com/showthread.php?18528-NP9870-S-Nvidia-980-Overheating-nvlddmkm

Getting the feeling this GPU just flat out does not belong in a laptop.

 

Was the heat issue resolved by repasting?

Also, your black screen sounds like driver problem. Does this thread help - http://forum.notebookreview.com/thr...g-drivers-mirror.786094/page-10#post-10214806. Especially part about DDU.

The only other thing that comes to mind is a loose connection in a cable to the panel in which heat causes the display to lose its signal. Do you have an external monitor to check to see if the card still produces a signal when used but hooked up to an external display?

 

Unfortunately the repaste didn't solve it. It did help a good bit as the initial paste job was done rather poorly.

I've tried about 5 different drivers. Starting with the ones that came with the system. I now have the latest beta drivers installed, but issue persists. It consistently crashes at around 70c. I didn't use DDU though, but I did use the clean install option on every install though. I had no issues with the actual install of the drivers. I'll try with DDU this time and see if that helps.

I think I can manage to dig up an old monitor to try, but would that cause the event log to throw errors? Any idea where I can check this cable (the literal entire bottom of this laptop comes completely off... quite nice actually). I remember having a similar issue with my previous laptop of which I had to RMA, but they never told me what was wrong with it. It's just unfortunately a bad time for me to be able to RMA right now. It seams like the GPU fan just isn't working very well at all.

 

Are you getting any errors in the event log when it crashes like that? Does one fan seem like it is working harder than the other fan as well? I don't think that would be the error for the crash, but would be something that we would probably want to look at as well if it isn't working.

I know the service manager is checking in to some of this for you. But if there is anything I can do to help out as well, feel free to contact me here or email me. We want to make sure we can find a way to get this taken care for you.

 

Yes, sometimes event errors are logged. Basically the driver stops responding (TDR). I've already adjusted the TDR delay which doesn't help and will be running DDU and installing latest beta to see if that can help.

One fan is absolutely working harder than the other. More specifically the CPU fan is working fine, but the GPU fan barely seams to move air. It's possible there's a bad connection with the fan or this fan is just no good.

On a side note as updated in my XoticPC topic it's extremely weird to me that the copper heat pipes have all been painted black. Does anyone else with this model have this issue? When you open the bottom see if the heat pipes are painted black or plain copper. Should be plain copper.

 

is this a common problem with this laptop? Why is it overheating at 70c? That isn't that hot for a GPU I have seen people who run their cards A LOT hotter than that and have no problems with overheating at all....even if it is a fan problem 70c isn't that hot and shouldn't be overheating at that temp...

*edit* I am almost positive all the new Sager/Clevo's heatpipes are black I think they did that because of people calling them "batman" lol but my laptop is "older" and my heatpipes are just plain ol copper and they look awesome

 

Yeah, the models I've all seen are all that way with the black heat pipes. I don't think if it was making a significant performance change in heat that they would have painted them. I'll let our service manager know about the fan issue though as well, so he can check in to that.

 

No idea if it's a common problem with Sager laptops, but seams to be a TON of Nvidia BSOD issues lately. So I'm not sure what to make of it. My HWINFO report dumps for both crashes are attached to my XoticPC topic if you think it could be something else failing.

Ah, ok thanks for the black heat pipe explanation, lol. Didn't realize they were doing this on purpose.. to components no one sees unless they open it, lol.

Ok, thank you!

 

Can you check the GPU fan cable to the header? Is it loose? If it is the header itself, a temp sensor or something on the mobo has gone caput, then you're most likely going to have to find some way to get it in to be looked at.

Another thought, with the P870DM, there was an EC problem that needed correcting in the BIOS. @Phoenix or @Prema may have more additional information. Since the EC has some decisions to make regarding fans, is there a way to check to see if you need to update the EC?

Also, perhaps @pat@XOTICPC or @Meaker@Sager may be able to provide some assistance as well.

----

p.s. - Can't get to any crash/dump reports due to Xotic's forum policies for unregistered users. If anyone on these forums asks for them, you'll most likely need to repost them here.

 

Even if something is wrong with the fan or its loose, the GPU shouldn't overheat or have problems at 70c maybe in the high end of 80c but it shouldn't be having problems at 70c.....

 

Agreed. To me, that part still sounds like driver related or perhaps something not quite physically connected with the GPU.

 

I'll check all the cables from the GPU when I open it up later today, but when I replaced the paste the connections all looked fine. The sensors are reporting the temperature fine it seams.

EC problem? Do you have a link to that discussion by chance? Do you mean the system BIOS or vBIOS? Current BIOS appears to be 1.05.03LS1. Below file contains my HWINFO system report and 2 sensor reports (both up until BSOD).

https://drive.google.com/file/d/0B6YuIA-KqT-dUkJlOXd6YkFDalE/view?usp=sharing

Sorry for the Google Drive link, but these forums won't let me upload attachments. Button is there, select a file, and it does nothing from there.

I agree, it shouldn't be overheating at 70c. Especially on a laptop, BUT this is supposedly the desktop GPU slapped into a laptop, which is kind of where my concern comes from. I am starting to think this is likely a driver/software problem more than anything, but the under performance of the fans is still concerning.

 

No. Not VBIOS, but regular BIOS. The Clevo BIOS consists of the main BIOS for the system and the EC portion. EC stands for Embedded Controller.

IIRC, what originally was discussed was a throttling issue in the CPU, but there may have been some implications on the GPU benchmarks. In any case, I believe this was fixed by Prema, and I believe Clevo was made aware of the problem. @Mr. Fox might remember where this was at, but here's the resolution :

http://forum.notebookreview.com/thr...ers-lounge-phoenix-has-arisen.781814/page-157

Now, I don't necessarily think this is related to the problem you're describing. And I have no idea if the GPU fan is controlled by the EC and limits/zones set there or perhaps by some other device. Regardless, I wanted to point it out just in case.

 

Hmmm.... Interesting post!!!

@Mr. Fox, could you take a look at what @Krileon is reporting. This seems very, very familiar - http://forum.notebookreview.com/thr...enix-has-arisen.781814/page-153#post-10150708

Take a look at some of the hints he has (or PM @Mr. Fox). Things like Control Center, NVIDIA GPU Scaling, etc. Maybe something there will help.

So it doesn't look like it is driver related after all.

 

Also, I did ask to see if we can get you the EC and vBIOS link sent over to you from Sager. Not sure if it would help, but couldn't hurt to try.

 

My 2 cents, for what it's worth...

@Krileon - I think you have a bad GPU. The log files show no problems that I can identify. Temps are fine. The problems I had with the screen going black with 980M on two difference laptops were caused by a GPU defect. In both cases one of the GPUs (in SLI) was messed up and causing black screens. In both cases this eventually, over time, lead to complete failure where the machines refused to POST or the screen was black all the time until I removed the offending GPU and ran only the good one by itself.

If this is happening with temps well below 80°C and no overclocking, it's time to get that GTX 980 swapped out for one that works correctly. The fact that there are not a bunch of other people with the same machine having the same problems you are experiencing is further indication that a repair needs to take place.

 

I agree at this point an RMA likely should be done. Problem is they want me to ship the entire system off, which is not acceptable at this time. I'm in discussion with them regarding putting a temporary hold for the amount of a new card on my bank card while they ship a new one out for me to replace then ship the old one back. It's just not ok for me to be without my system for 2 weeks when I also work from it and it has several thousands worth of sensitive data that frankly I don't want some random employee ever having access to. Replacing the GPU is a trivial task for me so I'm hoping this won't be an issue for them to do.

Also regarding your "Also, look in the BIOS and if the option is present confirm NVIDIA GPU Scaling is set to the "Disabled" option." comment in the above linked topic in my BIOS GPU Scaling is indeed enabled. Should I disable this?

Edit: Ok, my BIOS more specifically says "GPU Performance Scaling".

 

No, you can leave that enabled. I misunderstood what @Prema said about it and he meant the opposite of what I said. Sorry for the confusion. Enabling it allows overclocking. Disabling it is intended to block overclocking. If you are using Prema firmware it is irrelevant, since he circumvents their evil intentions. As far as your black screen issue is concerned, there is certainly no harm in testing it both ways as a temporary workaround. I doubt it will change the problem, but nothing ventured, nothing gained, and nothing lost either way.

 

Ok, I used DDU, installed 364.51, disabled GPU Performance Scaling, uninstalled Sager Control Center (Hotkey), and enabled VSYNC. Lasted about 20mins before BSOD. HWINFO dump shows it crashed at 69c. Keeps doing it around 70c. Could there be something wrong with the BIOS thinking it's overheating? Attached is the HWINFO report from this test.

 

Possible, but not likely. A weak component on the GPU may be getting too hot due to a problem (defect). The only sensor we have is the core, and your core temps are fine. Most likely cause is a defective GPU or bad MOSFET(s) on the GPU. One of my bad 980M cards was fixed by replacing the MOSFETs. I was able to avoid black screen problems keeping it super cold on AC for a while, but eventually even that was not enough. It was out of warranty, so there was nothing to lose. New MOSFETs stopped the 70°C black screen problem on that GPU. The other one was sent back to Eurocom. @ssj92 also had defective 980M cards. Your GTX 980 is the first one I know of that is messed up like that. There may be others I haven't heard about. Nothing is 100% failproof, but QC on 980M seems to be a lot lower (a subjective opinion on my part, based on my limited experience--no data to back it up) for 980M than it is on GTX 980.

 

Ok, thank you. I've emailed my findings to XoticPC. Hopefully something can be worked out to just RMA the GPU as i'm out of ideas on what more to try. Pretty much ruins my entire week and weekend on The Division release week, lol.

 

I'm not sure I've seen an instance where DDU actually helped someone resolve their issue.

Although you've contacted Xotic for help/RMA, do you have any minidump files you (or we) can analyze that might shed a little more light on things?

 

@Krileon - another question is what Geforce drivers are you using? The driver provided by Clevo or something else? A fair number of people are having issues with newer Geforce drivers being super buggy if not downright unusable. If you have not already done so, try using an older driver. Go to NVIDIA.com and select your GPU and OS, then search for drivers. In the list of drivers that appears choose the oldest you can find with support for GTX 980 notebook GPU and see if the situation changes. Try several drivers from the oldest forward. If you have the same behavior regardless of what driver is used, that would be useful information. To @Prostar Computer point, the minidump files might be useful. The bugcheck code may help pinpoint if something other than the GPU may be causing behavioral issues.

 

There's no minidumps. There's not even an event logged in event viewer. It just goes black and doesn't come back. I've let it sit for a good 10 minutes, but still no luck. It fails to recover every time

I've tried about 10 different driver versions. I started with the drivers included with the laptop, latest from sager, latest from clevo, then used the drivers from nvidia.com directly. They all have the same issue.

I just now tried lowering the game settings then watched the temperature with HWINFO. I can repeatedly confirm that as soon as it hits 70c it begins to fail. I kept tabbing out to drop the GPU usage and was able to play for nearly an hour by forcing the GPU to keep scaling down when at desktop to keep the temperature down. Something is wrong with a sensor or something it seams. There's no reason for it to fail at 70c as if it thinks it's overheating.

 

the GPU needs to be replaced, 99.9%. RMA it and save yourself from frustration.

 

I agree, but at this point I'm at the mercy of XoticPC to get that ball rolling. I've been in contact with them since yesterday regarding an RMA and keeping them up to date with my findings. I'm hoping the RMA for the GPU can get started before this weekend so it can get shipped out for next week or something.

 

Yup, totally agree with you.

I think that is pretty conclusive testing. The symptoms you describe are more or less identical to what I experienced. Once the screen goes black it never comes back again. You have to do a hard/forced shutdown holding down the power button. The 70°C core temperature may indirectly correspond to the point at which some other component fails or reaches a much higher temperature that we have no sensor reading for. It could also be the temperature at which two faulty soldered contacts separate and cause the black screen problem. I suppose the speculative list of reasons could be a long one. The magic number for both of mine was also 70°C. The only thing that was conclusive in my case was replacing the MOSFETs did the trick. Even after being fixed, while that GPU worked flawlessly at or near stock clock speeds and voltage it was nearly worthless for overclocking. It could not handle anywhere near what a properly functioning (ordinary) 980M was capable of handling in terms of core overclocking capacity. The memory overclocked fine, but not the core. A CPU or GPU that cannot be overclocked well beyond stock specifications is utterly worthless and a machine like that provides zero pleasure in ownership for me. Might as well save a few bucks on a BGA turdbook if it cannot hold its own in overclocking when pitted against otherwise identically configured systems.

 

Decided to take another look and completely pulled the GPU out. There's dried paste on components no where near the heatsink (even on some of the soder). I don't know what to make of this. I cleaned it up and repasted it, but didn't help. I agree, it's shot and on its way out. I'm hoping the RMA process can get going tomorrow.

This would be my second laptop in a row from XoticPC where the GPU is basically DOA. It's a bit discouraging to say the least. Already passed my old one down to my wife and get to watch her play The Division while I sit here with an expensive paperweight, lol.

 

As long as it's non-conductive paste, it's okay. It's worth making mention of that when you get your RMA created, though; bad paste jobs are bad for temps.

 

Ok, some more findings. I installed the clevo vga drivers below (nvidia 354.35).

http://www.clevo.com/en/e-services/download/ftpOut.asp?Lmodel=P870DM&ltype=9&submit=+GO+

I no longer get a black screen, but the video driver does fail and windows does manage to recover from it. The game obviously crashes, but the computer doesn't get stuck in a non-recovery state. This all started to happen around 70c (which it hit then quickly dropped back down). Below is the event logged.

Code:

- <Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
- <System>
  <Provider Name="Display" />
  <EventID Qualifiers="0">4101</EventID>
  <Level>3</Level>
  <Task>0</Task>
  <Keywords>0x80000000000000</Keywords>
  <TimeCreated SystemTime="2016-03-10T23:41:09.753703800Z" />
  <EventRecordID>9983</EventRecordID>
  <Channel>System</Channel>
  <Computer>Kyle-PC</Computer>
  <Security />
  </System>
- <EventData>
  <Data>nvlddmkm</Data>
  <Data />
  </EventData>
  </Event>

Attached are some screenshots of the artifacts that begins to appear around 70c. I've also attached the HWINFO report of this crash. I noticed my GPU maximum power in HWINFO hit 202.723 W, is that normal? This log also much much better shows when it crashes as you can see the GPU load suddenly drop to 0%.

Update: Retried Sagers (354.35, same as clevos; odd?) below for the heck of it and it gives black screen. So far Clevos has been the most stable (windows was able to recover). Going to try a few more I guess.

http://www.sagernotebook.com/drivers.php?cat=627

Update: Tried oldest available nvidia driver (358.50) and same issue with black screen. I don't think it's a temperature issue, but entirely due to being at maximum load. I compared the GPU core load on each of these and they're at 99% when it failed. I think this basically confirms bad GPU for sure.

 

Whoa, that's really messed up... I think any testing you are doing now is only going to add to your frustration level. Even if you find a driver that mysteriously cures your woes it will only be a temporary fix, as the clock is ticking and the Grim GPU Reaper is coming for that one... soon.

 

Was bored with nothing else to do so figured might as well try some other drivers. Didn't work as I suspected it wouldn't, but better than doing nothing I guess, lol. Yup, it's getting close to death so I'm not going to push it anymore and won't be gaming on this system until the new GPU is in.

 

Well, last ditch effort I underclocked the core by 400mhz and it hasn't been able to hit 70c. Highest it hit is 66c. No crash yet, lol. 70c must be a magic number. I know this is just delaying the inevitable, but if it can last until the new GPU comes in then that'd be great, lol. Anyway, thank you everyone for all your help. Hopefully have the RMA process going tomorrow.

 

Some other component on the card must be overheating. Maybe there's a thermal pad missing somewhere. You should take a pic of the heatsink thermal pad layout to show XoticPC if it's right.

 

I know we are checking in to it and would be able to RMA it. We are working with Sager to see if we can cross ship the GPU to save some down time. I would agree, I think you have provided plenty of testing info at this point as well and can relax on that point until you hear back on the RMA. Hopefully you can keep it under 70 for the time being to enjoy this math game you are talking about.

But I'm sure you'll hear back from the service manager as soon as he gets answers. Feel free to reach out to me as always if I can help.

 

Well, took the photos as @ssj92 suggested. They're attached as follows. You'll notice in one of them I took a closeup of a component that looks not so right. I'm not sure what it's covered in, but looks like paste. I didn't downscale them so they can be zoomed in for further detail as needed.

 

That looks to be the bios chip and thermal paste to me. I assume you recovered the core?

 

Yup, and it wasn't the only place that thermal paste was just sitting all willy nilly. It was dripped all over the place. I cleaned most of it off, but couldn't get all of that off that chip. Yes, I applied new paste to the core.

 

Any change in performance once you did that?

 

No, the card is clearly faulty. At this time I'm just buying my self time hoping it won't fail before a new GPU arrives. Downclocking to keep it under 70c is basically my only solution discovered, but it's obviously not acceptable for a brand new laptops GPU performance to have to be cut in half to function, lol.

 

Yeah, I'd agree, it isn't something we like to see at all either.

 

Good news! RMA is on its way. Hopefully next week things should be running good. Will report back my findings and tests after the new card is in. Also ordered a new tube of MX-4 instead of using the paste that came with the computer.

 

MX-4 is junk.. Cancel your order and get IC Diamond or Gelid GC Extreme.. much better and only slightly more expensive..

 

I would certainly not call MX-4 junk. I have used it on tones of laptops, desktops, etc. It keeps my 5960x @ 4.3 GHz pretty cool, and even my laptop CPUs and GPUs cool. Thermal pastes are only a couple degrees apart from best to worst: http://img.techpowerup.org/120126/thermal.png

http://3.bp.blogspot.com/-Hd2g3UaTweM/VLU5ibAbN4I/AAAAAAAAKgY/zagwpzmXC2I/s1600/temp+diff.jpg

http://media.bestofmicro.com/6/F/396807/original/01-Water-Cooling-High-Pressure.png

I feel MX-4 gets a lot more hate on these threads than it deserves. I never had any issues with pump-out or anything else.

 

This is not entirely true... I have tested different brands of high end thermal greases on my OC'd i7-4930Mx Hotwell in my AW17. You can not use thermal grease rewievs as a 100% truth if they are done only on desktop processors/test bench and not on laptop hardware with different cooling. You can look at several rewievs of thermal paste... Same paste can give completely opposite results done by another tester. There are enormous differences in cooling between laptops vs. desktops. 1-2 degrees maximum temperature difference as you say between good paste is nonsense on laptop hardware. The temp difference between thermal paste's is bigger for laptop processors vs. processors in a desktop. The difference will be bigger the more you push hardware to the max. You must neither forget the life span. I can kill a highend thermal paste under one week, if I Oc my Hotwell extreme and run bench all day. Only Liquid ultra is useble on that laptop processor.
Look at one of the pictures with the test of Gelid Extreme you posted. Gelid Extreme is torn apart by AS5. This is just nonsense... All tests of Thermal paste show different results. Some tests show that Liquid Ultra only gives 2 degrees better temperature than the best standard paste. This is also nonsense. The real difference is 10 degrees or more on laptop processors (OC'd) in bench. This is well proven by me and a lot of others. It's the reality that counts. Not thermal paste rewievs done on test bench/desktop processors and with different colling.

 

Depending on the heatsink MX-4 is a little more prone to pumping out, usually IC diamond and the thicker pastes get along better in a mobile environment.

 

If I remember correctly... A few people here in the forum has posted thermal paste results that are startling. With differences of up to 10 degrees between 2 quality thermal paste brand on their laptop processors. Application of both paste brand was done properly by the people.
In so-called test rewievs done with the same paste's on desktop processor, was the difference in temperature only 1-2 degrees... So you can not say that the difference is maximum 2 degrees between different paste's on laptop hardware who have a more fragile cooling...

 

he didn't

 

Just look at how fraudulent such rewievs of paste are (see images of two tests). Gelid Extreme get a totally different result against AS5. This difference in temperature result is not sensational. It depends on who performs the tests and the use of different hardware. This is like the lottery. You have to try different paste's self to see the difference.

And look below ↓↓↓. Gelid Extreme is nearly 5 degrees colder than AS5.

 

I don't think they are fraudulent or fake like you are claiming just results vary...