GPU Overheating, but not what you think ~

Hi,

Since I got my m15x in early September it's given me nothing but worries and grey hair - if there's anyone here that can help with my problem I'll be forever greatful - so far the tech support people at Alienware have been very understanding and patient with my issues, and I praise them for that, but the latest problem with the system is unable to be resolved - let me explain...

In the simplest terms, the problem is that the GPU Fan sometimes doesn't kick in when I believe it should, causing the GPU to overheat until the PC shuts itself down as a safeguard.

I know that the fan works, because as soon as I start up the PC again after it shuts itself down or I shut it down, the fan kicks in full blast, blowing out all the heat that has gathered up. I don't know how hot the GPU gets at its hottest, but using two monitoring programs I saw the GPU temp raise from 55 degrees C to 122 degrees C after about two minutes of gameplay, without the fan even activating.

During gameplay, once the GPU hits a certain temp, the framerate drops to about 5fps, and remains at that until the system shuts itself off.

Now I was under the impression that downclocking was occurring, and after reading about the m15x overheating issue I figured the BIOS update would fix the problem, raising the fan rpm. BUT, as I discovered, downclocking was not occurring - the GPU was simpy getting too hot without the fan - I flashed the BIOS to the newest version - X36, but the problem continues (though X36 did fix the popping sounds in the Audio, however).

Now if anyone has any idea what the heck is going on I would love to hear it - the solution the AlienWare support guy gave me was to replace the fan and heatsink, which I think is ridiculous, especially since the fan does work fine during startup. My suspicion is that the problem is in the heat monitoring of the GPU, but I'm no expert in this matter...

So please, if you know anything, please let me know - I'm desperate. Just today the GPU temp started rising from 55C - 108C without the fan activating during regular non-gaming usage before I freaked out and shut it off...

Bests, MB

 

I agree with you on the heat monitoring. It should automatically throttle down and turn the fan on. Call them and try to discuss it with them...it may work. Or, you can always try the demand approach until they get tired and give in and replace the gpu unit.

 

perhaps getting a clevo m860tu might solve your problem.

 

Hehe, good advice

Could be a broken fan. Is the fan running at all?

 

I have spoken with them, quite a bit actually. Probably spent about 3 hours on the phone today talking to 3 different agents, all with different opinions.

One guys said there was a heat monitoring bug in BIOS X36 that they're current'y working on - but wouldn't this then be happening on all systems with X36? He also said my graphics card wasn't meant for gaming (I have the quadro 3600M), but that's aBS answer cause even a 3600M shouldn't overheat to 122C!

ANother technician had me reseat the fan connection, which was ridiculous, but I didn't and it didn't change anything.

So finally I agreed to have them send me a new heatsink and fan, though I doubt it's the problem...

Isn't there a place to manage fan speeds and temps, and force the fan to go on if the temp exceeds a certain amount?

Bests, MB

 

Thanks

I definately would have done that and returned the m15x within the 30 day grace period had it not been for the 15% restocking fee Alienware charges on returns. My m15x was $5200, and I refuse to pay $780 for a restocking fee. So it looks like I'm stuck with the m15x - and any help regarding the problem would be, well, helpful

Bests, MB

 

Yeah - as I wrote above, the fan does work at varying speeds - it's just that sometimes it doesn't go full blast when I feel it should - in fact, sometimes it may stop all together, causing the GPU temp to rise exponentially.

But it is always the case that as soon as I reboot after oneof these heat increases, the fan goes on full blast and cools the GPU back down...

Bests, MB

 

Sounds like the motherboard sensors could be going out of whack. But, wait for the new heatsink and fan to come first before doing anything else. You can try flashing your BIOS back to the X32 version to see if it helps any.

 

I'm thinking it might be a sensor problem too, and one of the tech support guys said that X36 had a bug regarding temperature, though I think what he was saying was that the BIOS may be thinking things are hotter than they are, and in my case this is opposite.

Additionally, I first started noticing this happen before I flashed to X36, and I'm not sure what version I had before I did the flash - I received it towards the end of august - I don't know what BIOS version was being shipped then, but maybe someone does...

Bests, MB

 

Well, if you are using apps to record temps, then it may not be a hardware issue. It sounds like a bios / graphics card issue to me. You could have a defective card but i dont think your mobo sensors are messed up if they are returning correct values. Let us know what happens if you try to flash back to x32!

 

Well, the two 'apps' I'm using to check the GPU temp are 'hwmonitor' and 'gpu-z', and they both show the same temperatures. It's really hard for me to pinpoint the problem, especially because so many people are saying different things.

From what I can tell it can be the BIOS, the GPU, the fan, the motherboard, and who knows what else. But I doubt it's the BIOS since no one else is experiencing this after flashing to X36, and it was happening to me both before and after I converted. I doubt it's the GPU since it appears to work perfectly fine until it overheats. I doubt it's the fan because when I reboot after an overheat the GPU fan goes on full blast, effectively cooling down the GPU from the previous overheat. And I doubt it's the sensors because both 'gpu-z' and 'hwmonitor' seem to be able to monitor the temp just fine. But this leaves me with no clues - am I missing something? All I know is that the fan doesn't kick in when it should, but this can only be observed occasionally.

Now I'm curious, I just flashed to X36 BIOS, but I'm not experiencing any of the loud fan noises people are talking about - at all - the fans appear to be acting as before, with the same overheating problems. Could someone please describe what their fans do after installing X36 - are they on full constantly or only sometimes?

Bests, MB

 

After flashing to the X36 BIOS, my fans starts blasting as soon as I turn on my notebook.

The fans (at least for the X34 and the X36 BIOS) are based on a power setting basis, instead of a temperature basis to resolve downclocking issues experienced in earlier BIOS:

The reason why the fans may not be running while you are on the computer may be because you may:

#1. Have Stealth Mode enabled.
#2. Are using a power-saver setting / any variation of energy-efficient settings.

Seeing as how Stealth Mode effectively lowers the fan speed even during the boot up process, we might be able to say that it is not stealth mode that is causing the problem but rather the OS (or any software on the computer) particularly because the fans run fine during the boot up process and not while you're into Windows already.

 

good luck with that man
im sure they will change it within a heartbeat though since dell owns alienware and nvidia promises dell money for every defective card.

 

if it was getting over 105c, then i bet your card is now fried like french toast. Seriously. DEMAND that they replace the vid card and probably the motherboard. almost sounds like a voltage issue to me coming from the PCI-xpress slot.

 

How would I know if my card or motherboard was fried like french toast? I am still able to run games - there just seems to be improper management of the fans in my case - check my thread below fro my latest case study ~

Bests, MB

 

I just recently 'observed' and tested my system to try to isolate the issue, and this is a step by step walkthrough of what happened:

*NOTE: I just got my computer, and haven't had any time to install any third party software except for a few games, so I'm fairly confident that no software that I've installed is causing this issue...

-I booted up my computer and started HWMonitor and watched the temp for about 10 Mins. It looked like it began around 50C and started incresing slowly. This is without starting any other programs.

-Once the temp hit 68C, the fan began to spin, blowing warm air out the back. If it was spinning before this there was no way to tell as there was no sound or no warm air. This new spinning was not full power - just gentle blowing.

-The temp then began to lower, till about 55C, when the fan went silent. After this the temp began to increase again, till about 70C, then the fan went on gently and cooled the GPU to about 55C again. This happened about 4 times.

-I then started up a game (Lost:Via Domus), and ran it at 1280x720 with lowest graphic quality for about an hour. Every 5 mins I would alt-tab to windows and check HWMonitor, which actually stayed between 58C and 66C the entire time. I thought this was curious as before I had only managed to run the game for 2-4 mins before the framerate dropped to 5fps and the GPU overheated.

-So anyway, thinking the problem had vanished magically, I closed out of the game. HWMonitor showed the temp at 60C. And then I watched it to see if anything happened. At this point the fans is completely silent. Sure enough, the temp began to rise, faster and faster - 61, 62, 65, 71, 78, 84 ~ even though I wasn't running any graphical applications (besides windows), and the fan was still silent.

-But here's the weird part. Once the GPU had reached about 88C, I touched the stealthmode button to see what would happen. And lo and behold, the stealthmode button dimmed, and the gpu fan kicked in (I would say at about medium speed), thus lowering the temp back to around 62C. But not once have I witnessed the gpu fan go on full blast after flashing to X36.

-At this point I thought I had figured something out, so to test it, I immediately started up the game again and let it run for about 3 mins. I then alt-tabbed to windows, and the gpu temp was at 79C, and the fan was silent.

-I touched the stealthmode button, dimming it (and turning on stealthmode?), and the fan kicked in. I immediately touched the stealth button again, and the fan remained on, continually cooling the gpu to about 65C.

So it's almost as if I was slapping the fan in the face when pressing the stealth button and saying 'hey, pay attention!'.

Does this make any sense?

Also, the hottest I've measured the GPU at was 122C, and I'm sure it's been hotter since it's shutdown twice automatically as a result of heat. How would I know if something's fried?

And to be clear, when you press stealthmode and the button dims, you're turning stealthmode on right?

Bests, MB

 

So far I have never used stealthmode, and I always have the AC adapter plugged in. I also have the power setting set to High Performance. When windows starts up, my stealthmode button defaults to bright, and I assume that this means that stealth mode is off...

Bests, MB

 

Eh, I kind of had a feeling it was Stealth Mode. When the stealth mode button dims, it means stealth mode is turned off.

Stealth Mode attempts to conserve energy by downclocking your GPU and your CPU, turn down the fans (to a minimal level so it's silent) and lower your brightness settings (though you can always manually adjust it back to full brightness even with stealth mode on).

I suppose now that you've figured out the issue, the culprit was in fact Stealth Mode. You had Stealth Mode on the entire time, my friend!

Now that we've isolated the issue, I have further questions to ask you:
#1. Can you figure out if stealth mode turns itself on every time you boot up?
#2. If not, does it turn itself on after several minutes of gaming?

If you can answer 'yes' to any of these two questions, try uninstalling and then reinstalling the Alienware Command Center. See if that addresses the issue for you.

 

Ok, I really don't wish to overheat my system intentionally with again unless completely necessary. Luckily I did take a screenshot of cpu-z when my ssytem was at 117C which I'll attach here.

It may also be good to note that Speed fan doesn't register the speed of either the cpu fan or the gpu fan - the box is simply empty.

To answer your questions:

1] I first started noticing this problem after the system was returned to me from aw repair (my lcd screen had a bad connection so they replaced the wire). This is approximately 5 days ago. But the first time it happened was during gameplay - the framerate slowed to 5fps and the pc shutdown automatically shortly thereafter.

2] I always have stealthmode disabled (when the button is set to bright). Strangely enough, as I described in my post above, pressing the stealthmode button seemed to trigger the fan to go on, and the fan would continue even if I turned stealthmode off again.

3] It really depends what I'm doing. Lets say this faulty mode is called 'bad-fan-mode'. If I'm doing nothing (just staring at the desktop), and the computer goes into 'bad-fan-mode', the temperature with increase gradually, one degree at a time. I watched this together with a tech-guy earlier today, and it went from 55C to 118C without triggering the fan - I then rebooted the system, and the fan blasted as soon as the system started up again. BUT, if the system goes into 'bad-fan-mode', and I'm in a game, the temp can raise from 55C to 125C++ until the system shuts down, since the fan doesn't activate properly...

I hope this helps you help me

Here's the screenshot - it's of cpuZ the first time I tested the heating issue:

[I'm not allowed to post images yet, so here's the semi-link]
masterbjorn.com/shared/overheated.jpg

Bests, MB

 

Ok,
When the touch Controls are Bright that means they are ON. If they are Dim they are OFF. You want stealth mode to be off. Just put the stealth mode dim, see if that helps. That is why its getting to hot and the FPS are low.

 

Okay, issue resolved for now. We'll need to wait for further testing to confirm that what I've walked him through last night has permanently fixed the issue. Initial testing after the resolution has shown that the fans are up and running as they should be.

 

what'd you do?

 

I've walked him through "resetting" the touch keys by uninstalling and then reinstalling the Alienware Command Center. I figured that Stealth Mode was turning itself on every time his computer booted (or whenever he exited a game). Since the fans were blasting away during the initial boot up process, we can probably infer that it is not a hardware related issue; the fans aren't broken.

I have had several instances where I've personally had the bluetooth turn itself on and off whenever it felt like it, the Alienware Command Center opening itself constantly, and the volume slider flicker * (and therefore adjust the sound by itself) and since all of those issues can essentially be resolved through simply uninstalling and then reinstalling the Command Center (and therefore have it re-update the multimedia board firmware), I figured that this is, in a sense, like "resetting" the touch keys.

Having Stealth Mode on while he was gaming has caused the GPU to overheat because the fans are lowered (sometimes even completely stopped) in Stealth Mode. Because the GPU temperature has reached (and even surpassed) threshold temperatures, his computer reacts by abruptly shutting itself down.

Now to resolve this issue, we would basically have to reset the multimedia board (touch keys) so the touch keys would not activate itself whenever it felt like it.

Keeping Stealth Mode from turning on by itself would allow the fans to run normally, which in turn allows for the cooling of both the GPU and the CPU, which prevents the notebook or any of its parts to reach threshold temperatures, which prevents abrupt emergency shutdowns, which then would resolve the entire issue.

Like I said earlier, we'll have to wait for further testing to see if this has fixed the issue for him. Each system reacts differently to the solutions that are put out there and this may not entirely fix this issue for him. We'll have to wait until he finishes testing it out.

 

Yes, last night I received a great deal of help from Oceanus - more, in fact, than I could have expected. Most of the AW tech support guys I spoke to said there was a problem with the heatsink or fan, or claimed that my 3600M simply wasn't as powerful as the other cards and was therefore overheating (meant for CAD software, not gaming?) - But I had a feeling it was a software issue all along, and Oceanus helped confirm my suspicions

I owe him a thousand thanks and loads of gratitude for being patient and extremely helpful during the course of the fix - I was being very cautious and double checking everything he said, so it may have taken a little longer than standard - better than making small pointless mistakes and retracing your steps I suppose

So basically, I've been able to test the system now after re-installing command center (I played for about an hour), and the fans would go almost non-stop on medium, boosting up every now and then to high and keeping the GPU at a stable 70C-80C the entire time (according to ThermoFan).

I did notice now, though, that the laptop in its entirety got a lot hotter on the entire outer surface, moreso than before. I'm guessing that the constant flow of the fans essentially makes a heatsink of the entire laptop - correct me if I'm wrong.

So, for now the system seems to be in working order - it was quite a tricky problem and I'm baffled that Oceanus caught it on the first try. I'll make updates should anything change for the worse...

Thanks Again!

Bests, MB

 

I do what I can to help.

 

SO, just as an update, I think it's worth mentioning that I've now been able to play for several hours without the card overheating (thanks to the fix of Oceanus)

Also, according to ThermoFan, my GPU temp hasn't gone much above 85C during gameplay, and seems to idle in windows at about 70C.

Now I'm still a little concerned that my GPU got damaged somehow during the times when the GPU temp rised to 120C++ (before the fix). I'm not seeing any symptoms as of yet, so does that mean that everything is OK, or is there a possibility that those excessive temperatures could have caused damage that might prove to be fatal in the long run?

I got the 3 year extended warranty so I should be covered if anything does happen - I'm just curious. Is there an easy way to check if any damage has been done to the GPU?

Bests, MB

 

i think the first signs of damage would be video glitches, such as random characters/blocks of color/lines appearing on your screen.

 

http://www.techpowerup.com/atitool/

that will help test for artifacts... which is one of the early signs of permanent GPU damage due to overheating and such.

 

I've heard of ATITool, but I'm running Vista so that's not really an option. Luckuli I haven't noticed any lines or artifacts, so I'm guessing I was lucky this time (knock on wood)

Is there anything similar to ATITool that works on Vista?

Thanks!

Bests, MB

 

http://www.nvnews.net/vbulletin/showthread.php?t=97184

 

Hey, Thanks!

Bests, MB

 

AtiTool works fine with Vista sir. I use it all the time.

 

As an overclocking utility, it won't void the warranty will it? An agent once told me that overclocking either the CPU or GPU voids the warranty

Bests, MB

 

course it voids your warranty.

 

But, they don't know that. If you send it in, remove the program.

 

Hi oceanus I dont have a gpu unit overheating but I was reading this thread and this part about your bluetooth turning on and off whenever it felt like caught my attention. Because my m15x also turns the bluetooth on and off whenever it feels like, and after it goes haywire turning itself on and off like 20 times within one minute the bluetooth locks itself on and locks my entire touchpad center up. This does not happen all the time, usually once or twice a day after running for 24 hours, but it happens most often after I start up from sleep and hibernate. Usually it just happens randomly and for me to unlock the touch panel I would have to shutdown and turn the laptop back on.

I have tried multiple reformats to fix this but to no avil it never fixes. I have yet to try the new Alienware CC but I have reinstalled my Alienware CC multiple times also to no avil. Is there anyway to fix this minor annoyance?

 

When it turns on and off, it would show it on the screen right? Something like this?

If it does, uninstall the Alienware Command Center and also the OSD Utility, restart, then reinstall the OSD Utility and then the Alienware Command Center in that order.

I have a few other ideas on how to resolve the issue, but I need a bit more information before I can come to any conclusions. Just try my suggestion and we'll carry on from there.

Oh! And lastly, may I ask what Operating System you are using?

 

Same Problem! i recieved my area 51 m15x last december after 2 months of waiting until now i cant use my notebook to render still images & run animations.

when i start to render even small scene & set to high quality output & after 5 to 10 minutes my system restarts. before that i monitor the pc with nvidia system tools, thermo fan, HW from cpuid, the cpu reach to 100% with maximum temp of 54°C, & my Fx 3600m temp up to 74°C & the fan's RPM "5300 for CPU" & only "2200 for GPU" & BANG!!! restarts the system, i know that the system sdown in order to protect it.

I call the tech support (very costly to me) they check the system with temp monitor, & download the 3d mark06 & he let me finish the download coz it will take almost 1 hour. I installed & run the software with my GPU temp allready @ 68°C, the result was my system restart automatically & i wait untill my system will cooldown & start again this time its was succesfull but the result is below standard.

Hope if there is someone can help me, thats very fantastic.