M17x R2 - high temperatures after repaste

So I followed the "repaste" video guide linked from these very forums and did a repaste of my GPU's. Now my my temps have all gone to hell, sometimes as high as 120C.

I am guessing I didnt use a high quality enough of the paste...so I will go up there tomorrow and buy the good stuff (IC Diamond). Is there anything else I am missing?? Cause this is really throwing me for a loop...I thought I did a pretty good job.

Also, right after the repaste, I tried playing Crysis 2. There was a thin blinking vertical line on the left side of the screen and the screen would jump (like the vertical syn was out of wack). I was assuming this was due to temps being so high, does that sound right? Or if not...is there something else I could have messed up??

I really hope I didnt damage anything...dont feel like spending 2k on a new laptop right now.

 

Hold on the the purchase, whatever thermal paste you use it shouldn't lead you to that temperatures.

Re-check and redo the paste job, is more likely to be a bad application. Apply an uniform layer of paste, distribute it with the help of a credit card like object. Pay attention on the pads of the memory, close to the gpu they can keep the heatsink to make proper contact.

Also, IC Diamond in general is more meant to be used on desktop cpu's that have their metal protection than on a laptop chip. that paste is abrasive, i wouldn't use it on my rig, no matter the gain in temperatures. I would go with MX-4 if you have to purchase new paste.

 

Odds are either you used way too much (unlikely), or you just seated something wrong, and it's not making contact fully. Unfortunately you'll have to re-do it.

 

I dunno...it didnt seem like way to much. I used about a pea sized bead and spread it evenly over the flat shiny metal surface as evenly as I could. It seemed like alot at first...but then remembering what it looked like in the first place I dont think it was too much.

I guess I will redo it (and this time used compressed air on the heatsink/fan blades as well) and maybe try using a little less.

The more I think about it, I am wondering if I just used to much...hmm I dunno I'll have to check it out this afternoon/evening.

 

When you do the repaste make sure you don't touch the metal die (oil from fingers are bad). Also, try not to spread it yourself. When I did my repaste I put a rice-sized amount in the middle and put the heatsink back on. Pressed lightly and screwed it in. Opened it back up straight away to see how it spread, and it didn't spread well. I repeated, and the second time it worked well and temps are great. Just an idea for another approach.

 

Just to clarify, you only reapplied the thermal paste to the GPU die, right? Did you replace the pads that were on the RAM chips? If you used the old pads, were they damaged at all during the removal/replacement?

If your idle temperatures are somewhat normal, and then skyrocket once you put a load on the GPU, then you've got some kind of a contact problem somewhere.

You shouldn't have permanently broken anything. The GPUs have a thermal limit where they'll downclock to prevent damage to themselves.

Unfortunately the others are right, you're going to have to get back in there and see what went wrong.

Can you link to the video you followed?

 

As I understand it, I just put the paste on the gpu die (the shiny metal square under the heatsink). But I would say I used significantly more then "a rice sized bead". More like a pea sized bead...which in my mind is alot bigger then a grain of rice.

I thought the Thermal pads were ok, but now I am going to get new ones anyway.

My idle temps are 45c in main card and 39c in slave card...which is 5-6 degrees better then it used to be...which is why I thought af first my work was done.

Ooops it wasnt a video, just a photoguide:

[PHOTOGUIDE] Cleaning/Replacing Thermal Compunds (GPU+CPU) on M17x-R2

Oh and to the other post...i dont think I touched anything with my fingers...I used the screwdriver to gently remove the dried up old paste or whatever it was...hope that was ok...

Oh and lastly, I did not do the retention clip mod...nor am I going to as lovely as this went LOL!

 

You removed the old paste from the card with a screwdriver?!
I hope not, never do that, use alcohol or acetone instead.

Another important thing is to tighten the heatsink screws properly, in X, not too lose and not to tight.

A lot of paste can be as bad as no poste, remember, less is more.

 

Yeah, if you used a 'pea' sized amount on the GPU die, you're going to have a hell of a time cleaning up all of the excess that squished out.

On a bare die, like our laptops have on both the CPU and GPU, it is OK to use the spread method. Put an amount about the size of a grain of rice on there, put a zip-lock or plastic wrap on your finger, and then spread a nice and even coat on the die. Set the heatsink on, and then tighten the screws in the order stamped on the heatsink. Go about 1/4 turn at a time until they're all tight. There's no need to 'Hulk' on them. Once the screw stops turning, you're good to go. If this still doesn't work, then there is almost certainly some issue with the thermal pads on the RAM chips.

 

Well tonight I am going to redo it.
Does anyone have any objection to removing the thermal pads and just using paste instead?? All my local suppliers dont carry actual thermal pads.

Or, one guy said he can get some...should I wait to redo it and order new thermal pads?

 

The only way you can use thermal paste on the RAM chips is to use the retmod, which you've stated you're not willing to do.

 

mine is similar specs as yours. used mx-4 and fugipoly 1mm pads. and followed this guide below

Alienware M17x Teardown & Cooling Modification | Tech|Inferno

http://www.innovationcooling.com/applicationinstructions.htm

at first, i placed a small 'pea' size right in the center but apparently that was too small when I tested it with furmark. redid my paste job using a bigger pea size and that worked because i saw that my original paste did not spread/cover the gpu. i did not do the retention mod, which is probably why i need to use a bigger pea size. my furmark test did not go past 83* after 15 min, which is okay for me.

 

Ok here is the update...

So I did the repaste and all went (for the most part) well. It was a little messy but otherwise all went well. The paste I used now is much thicker and non-abrasive and at first all seemed well.

I was getting 40c in primary card at idle and 34c in slave card. So I messed around with some games and everything was looking GREAT!! Temps never got over 60c...where as before (even before I messed everything up) my temps would go almost instantly to the 90's no matter what game I played. Which was my main motivation for trying the repaste in the first place.

So I am thinking all is well and proformance seems to be MUCH better, let me run 3dMark 11 and see my score. I was watching the boring videos and was noticing significant increase in fps in various places...but then my kids started fighting so I had to step away for a minute or two.

Well I come back and my computer is turned off. Oh crap I am saying to myself. I press the power button and and the whole keyboard flashes but nothing happens. Do it again, nothing happens, press a third time and againg keyboard flashes but no power up. Crap!

So I unplug the laptop, remove battery and do a power drain (hold power button down for 30 seconds...) and then put everything back together. Computer boots up and all looks well.

So once again I play crysis 2 for about 15 minutes with no problems. Temps never get over 60c!

So I am lost?? What happened when I stepped away from 3dmark 11?? Did it actually overheat and shut off? I have run the benchmark program on this very laptop at least 6 times and it has never crashed before...so I dont know what happened. Does the whole "press power button and keyboard flashes but no system power up" have a specific meaning? Was it all just coincidence?

I will try an extended gaming session later on tonight and see how my temps go...probably save my temps to the log file and hope I dont get a crash.

Should I tempt fate and run 3d mark 11 again? Oh btw, the laptop made no sounds or beeps when it did the funky keyboard flash no power up thing...

Ok so after 33 minutes of Crysis 2 (my god that game is boring...) at extreme settings my temps never went over 83c in primary card and never over 79c in slave card. And those temps were very shortlived...so I would say this is a major success for me, a drop over 10c on average, plus no extended periods of higher temps.

 

Temps are indeed good, they can be better but definetly an improvement.

Have you run again 3Dmark11? i would have a look into that shutdown.

 

Congratulations on the fix!

As for the random shutdown, it has also happened to me. Every time it happens, sure as the sun comes up, the blue LED at the tip of my power cord is off, meaning the PSU has tripped. Sometimes it is for no apparent reason, but most of the time it is because I had it sitting on the carpet or covered by something. I believe they trip thermally, so that would make sense.

 

If the system isn't tripping during game play, I wouldn't worry about 3DMark11. But glad you got it all fixed as far as temps go!