Alienware M18x R1 CPU GPU overheat temps at low loads

Hi,
Hoping someone can help me. I have been searching for a couple weeks but not finding any solution. Any help is appreciated.

Having high CPU temp spikes on low load (30%) with temp spikes of 40 degrees. CPU usage is all over the place with extremely high spikes just using firefox. At times I can see one of the cores taking more load than others. The temp spikes quick seconds then drops and repeats. This gets particularly bad if HWinfo fan control is used.

Also GPU SLI is disabled yet the disabled GPU1 temp is running 61c with with its fan running constant. The GPU0 being used is running 55c with intermittent temp spikes to 63c since bios update. Both GPUs and CPU are at default clock speeds. I have not gamed yet because of this issue.

Mouse also jumps across screen at times. Slow response from opening windows or typing at times (desktop folders or firefox window may open with white screen then fill in a second later, typing is lagging). Upper right keyboard by touch panel hot to touch.

My system specs and what I've done to date.

Alienware M18x R1
i7-2630qm
580M SLI bridged
NVidia catalyst and drivers v388.43
SSD drive and standard HDD
Win7 Pro

Changed from Dell A05 BIOS to Unlocked Bios A04 (temps did get a bit better was running idle near 70).
Repasted CPU and GPUs with HTNoctua-14 and .5mm pads.
Cleaned minor amount of dust off heatsinks and fans with compressor and soft toothbrush. (was surprisingly clean when first opened)
Running HWinfo for monitoring with EC monitoring disabled in app.
Lappy sitting on 5 fan cooler. Moving lappy around cooler to make sure its not blowing hot air back in vents no difference. Cooler fan control up, down no difference. Off gives noticeable heat increase at least 10 degrees.

Nothing stands out in device manager and all scans of hardware report no errors.

 

I am not sure what paste that is. I have not heard of it before and nothing comes up on a Google search. I have used NT-H1 before and it is one of the worst thermal pastes I have ever used. Otherwise, It seems like you have done all of the right things. Please try some IC Diamond, Kryonaut or Phobya NanoGrease Extreme thermal paste and see if the temps are the same. Since you are trying thinner thermal pads (a good idea as long as they are still thick enough to make contact) when you take it apart, look closely at the thermal pads to confirm all of them are being compressed. If you still have a single pipe CPU heat sink, see if you can get a two- or three-pipe M18xR1/R2 CPU heat sink on eBay. With the stock one-pipe heat sink I always had higher than acceptable CPU temps with the 2760QM CPU no matter what thermal paste was used. The system is old enough now that the heat pipes may be losing their gas charge and not cooling as well.

A higher CFM CPU fan (J77H4) is also available, and it does help. Notice the CFM on the J77H4 is 11.7 CFM versus 4 CFM.

 

Thank you for responding.

I have had this all of two weeks and am about to loose my sh*t, lol.

It is Noctua H1 paste that I used. I will re-paste with another. Of the three pastes you mentioned which is the better one? should I go with a 1mm thermal pad? The heat sink is two pipes. There was thermal pads on the pipes, which I removed. It seemed to me that pads on the pipes would block flow of heat dissipation. The fans look original. If the CPU thats in it runs higher than normal by design, should I change the CPU? if the CPU were to be changed is there one you would recommend?

 

You're welcome. Sorry nobody else had any suggestions, but maybe those other folks that know their way around the M18x have not seen your post yet.

I would say Phobya NanoGrease Extreme of those three. The best CPU options available for the M18xR1 are 2920XM or 2960XM, but I would try to get those temps under control first. If you can get a three-pipe heat sink, do so. Then if the temps are good you can look for an XM CPU and 11 CFM fan on eBay.

Leaving the pads off the CPU heat sink was a good choice. Even Alienware stopped using the pad on it because it caused problems.

Using 0.5mm pad on the GPUs is fine as long as they are making good contact. If they are, thinner is better. If they are not, then the 1.0mm pad would be better.

 

Will do. Thank you very much. Your help is very much appreciated.

 

.

 

I repasted with IC diamond 24 carat and also fixed the pads. I am still using Ahmeds A04 BIOS and HWINFO64. Temps did drop and all was good...for awhile. However I am having a variety of problems now.

I am getting a roughly 30 degree temp difference between the 580m card being used which is generally running at 35-45 idle (HW reporting) and card2 that is SLI disabled idling at 60 degrees. I am running PHYSX on card1. HW fan control is a one off, fan control limits are locked at 2000 and 3800, fan controls are set to auto custom sometimes the fans will come on high and then just stop 2 seconds later, sometimes they are not activating at all as they should.


As of this morning I only have control over 1 fan in HWINO the other two are greyed out.


I am also getting ACPI events 13 and 15. Event 13 with the battery in and 15 with the battery pulled. I have experienced 3 shutdowns, which may or may not have been attributed to hibernation being activated. I have since deactivated hibernation.

I am also getting cursor jumping and minor latency when opening files or windows.

All hardware scans in support assist, BIOS, sfc and minor benchmarks come back passing and nothing stands out in device manager.

Virus scans with adwcleaner, hitmanpro, malwarebytes, avast, superantispyware and esetonline come back clean also.


I am running Nvidia Cntrol panel v388.43

My question is...could card2 be going bad? maybe the temps are actually higher than HWINFO is reporting which may account for shutdowns and minor latency?

Is there a better version of the BIOS that may fix the acpi issue?

Is there a better version of Nvidia catalyst and drivers than the one being used?

can you point me to where I can reset HWINFO to have full fan control and not have a lockout limit.

 

Yes, you could be having a hardware issue. It is a small miracle that your 580M cards have lasted this long. That's pretty amazing. You might try removing the slave GPU to see if the issues go away. If things return to normal, that would definitely pinpoint the problem as the slave GPU. Normally, the master GPU dies first. So, if removing the slave does not fix it, try removing the master GPU and using the slave in the master MXM slot. Or, remove both and see if you have issues running off the Intel HD Graphics.

You can try uninstalling HWiNFO64 (confirm all folders are deleted) and use a registry cleaner to remove all of the supporting obsolete keys from it, then unzip a new portable or reinstall. I always use the portable version and prefer that. Enabling EC support increases fan control, but comes at the cost of some things like DPC latency and random glitches (freeze on shutdown was one I can recall).

 

If one of the cards are failing even though scans are all passing I know that could cause a TDP shutdown but would that also cause event IDs 13 and 15? or are the events more a software setting?

 

Also, do you think disabling card2 in device manager help to rule out if its failing in lieu of a teardown to get the card out?

I remember before installing the Nvidia 388.43 catalyst drivers device manager was not reporting Intel HD on board.

 

I am thinking more along the lines of physically removing the GPU/heat sink assembly and SLI bridge cable and setting them aside while testing. If you have an SVL7 unlocked BIOS you can manually change to IGFX mode to disable the MXM slots and switch to the Intel HD Graphics. If you use the FN key combo to switch to integrated graphics now (which will disable the MXM slots) does the system behave the same, better or worse?

 

I disabled in device manager, which did nothing. The temps kept rising. I ended up having to physically remove the 2nd card. Temps are about the same, 60 for CPU and 34 on GPU(haven't received the multi pipe heatsink for CPU yet). ACPI event 15 is still occurring lappy wont stay connected to power now if battery is in.

 

I dont have the SVL BIOS. I couldn't get my hands on it. So I am running the ahmed BIOS A04.

I am not sure which function key disables the MXM. This lappy is new to me, I assume it is the f7 key that has ID GFX on it, lol

 

Yes, that is the one. FN+F7. On the newer machine (Alienware 18) I think they changed it to FN+F5. Since I have not owned any Alienwares for a couple of years (sold all my stuff) I could not remember. I am not familiar with the Ahmed BIOS A04. Maybe he is one of the people that have been stealing, selling, and taking credit for SVL7's work like some people have been trying to do to @Prema for a while now.

If all of the menus are exposed, that switch probably there.

 

OK, I disabled and ran on Intel it stayed around 50 both intel and CPU was still getting event 15 on boot. I switched back to 580m temps about 58 and CPU 45, still getting event 15 and curious thing HWINFO is reporting GPU2 temp rising, but that card has been pulled.

Fans got stuck on high RPM in HW fan control, system hang on reboot with black screen so I had to hard boot to stop the high rev of fans.

Have I mentioned lately how much I appreciate your help? lol

 

No problem. Glad to help you any way I can. That is what friends are for.

Try deleting your HWiNFO64.ini file and relaunch so it creates a new one. Very odd that temperatures are being reported on GPU2 unless you are in SG mode with Intel iGFX and master 580M active. If that is the case, perhaps HWiNFO64 is seeing the Intel iGFX as GPU1 and 580M as GPU2. If running in IGFX mode with the 580M disabled, I am not sure why.

Can you post a screen shot of the event 15?

 

Deleted ini still reporting GPU2. I have added the screenshots of both event 15 and GPU2 in HWinfo. 580M is no longer disabled, I am running off it now.

 

Now on Intel with MXM disabled. It was indeed reporting intel card as GPU2. HW was reporting correct temp of intel card and intel was running hotter than Nvidia when not on intel.

But now that I'm running on intel after a few mins its temps seemed to have stabilized at around 48.

 

Try turning off EC Support (uncheck box) in HWiNFO64 settings. Close and restart. If you get the message at launch about EC support, select the option to disable it and check the box to not show it to you again.

You will only have like 3 fan speed steps with EC Support disabled, but it might improve the behavior of your system. I recommend trying that. It should also cure the system hang on reboot.

 

That did solve the ACPI events both 13 and 15. Thank you.

Ahhh gah and true to form, when you solve 1 problem others will arise....I put my battery back in to make sure no other events occurred, while no events did occur It is now switching between battery and AC every few seconds. It will not just stay on AC.

 

I cannot remember if the old version of Alienware Command Center had anything to "manage" battery charging or not. It has been too long and I have forgotten so much. I would check there. If you have any kind of Dell Support software installed, I would get rid of that. Their battery saver software causes silly behavior similar to that.

You can also go into Device Manager and delete the "Battery" part of the hardware tree, then reboot. That sometimes stopped the "Plugged In, Not Charging" issue and it may have some effect on the weird behavior you are seeing now.

 

Battery was reporting 100% on screen was cycling between on AC power and battery power. It started right after I removed GPU2.

In command center the only options I see listed for battery are low and critical options.

I unplugged and let the battery drain to 97% just to see what would happen then plugged into AC for charging as soon as it reached 100% it started cycling on screen again between on AC and on battery.

 

What you describe is exactly what I was envisioning, and that is generally a behavior produced by software designed to minimize battery wear. Some laptops have BIOS settings that do it, but I do not remember any of my old Alienware machines having that BIOS option. Clevo and MSI have a BIOS feature for that and I always disable it.

It could be something else, but I would look for and uninstall any software from Alienware/Dell that might cause it. If you have the installer file saved, you can try uninstalling Alienware Control Center software and see if that behavior stops after rebooting. If it stops, reinstall Alienware Control Center and hope it stays gone.

 

I checked the BIOS, but no function in there I could see that would affect this. Deleting the tree had no effect. Uninstalled support assist. Uninstalled command center had no effect.

 

I have tried a few things, to no avail. I did a power drainage a few times, didnt work. I rebooted on power only and then inserted the battery after windows powered, that did nothing either. I tried a sys restore as I had one I created yesterday, that did nothing.

 

I am out of ideas on that one. Not sure why it is doing that. Maybe the battery has an issue.

The only thing I can suggest for that is to run the Pre-Boot Assessment and see if it flags the battery as having a defect.

 

UPDATE-
All problems resolved. I ran diagnostics on the battery and power, checked wattage etc. all diagnostics came back clean.

I resolved the battery issue by changing battery settings in Command Center to not power down on Critical status, unplugging and letting the battery completely drain out, i put all intensive power sucking programs to run and then let the PC die. I then pressed and held the power button for 30 seconds, plugged it back into AC then booted up and let it charge to 100%

I watched it most of the night for heat and all the other issues or any new flags, so far all is good. I even got my game on for 6 Hours... :-O

Thank you for all your help Mr. Fox. You were a tremendous help on areas that were stumping me.

 

One final question. When you had an AW do you remember if SvLs A05 BIOS had good fan tables to keep the system frosty without having to use HWINFO for fan control? or is HWINFO my new reality for cooling?

 

Great work, Sister. Glad you got it sorted and glad I could provide some ideas to investigate.

No, the fan tables are in the EC firmware and hardly anyone knows how to mod it. If the mod fails it will brick the motherboard and recovery of a bricked EC is very difficult.