Hello everyone, I thought I would update everyone on my progress in reapplying thermal paste to this card. I was recently having problems with increased temps by about 4-6c and I couldn't quite pin point the reason. I was told changing the thermal compound might help, and was told by two different EVGA reps two different ways how to do it. (I went with the first one, as linus did the same thing for his 960 acx)
 
The process:
I want to start out by saying just how tight the four screws holding the GPU to the heatsink are, they are really in there and required quite a bit of force to unscrew. I made sure not to strip them, but they were in tight.
After the screws were off, I had a fair bit of trouble actually taking the heatsink off and the reason for that was how tight the LED connector was plugged in. It was also very hard to undo, and it is necessary for continuing.
 
Once that was done, I did not need to remove the fan header to continue, I merely opened it like a book and lay the heatsink on one side, and the exposed GPU on the other. I spent quite a bit of time using a coffee filter and 91% isopropyl alcohol to thoroughly clean the absolutely dried and cracked thermal paste that was already "applied" I then went with the line method for applying my MX-4 compound to it.
 
The aftermath:
Well, I installed the card back into my system and it was functioning, but there was little to no difference in temperature. Maybe 1c either way (Hotter or colder)
I'm unsure if I perhaps didn't apply enough thermal paste? Jayztwocents uses a lot more paste, in a way that EKWB recommends, and others spread it or just added a dot. (There's a lot of discussion on which is the "right" way)
I tested for quite a long time, maybe 2-3 hours continuous use in valley and got around 81-83c on a default fan curve maxing out at 62%. I got 77-79c on an aggressive fan curve at 80%
 
The Conclusion:
Overall it was a fun learning experience. I will admit I was a little intimidated to take apart a GPU I had spent 600 dollars on a year ago, but I didn't really gain anything by doing so. I believe anyone with an hour of free time, and some confidence could do this procedure, but in my case it didn't help any.
 
If you have any questions let me know, I can go more indepth about things if you'd like. I just felt like adding my experience to those who may be on the fence in doing this.

XrayMan
 
Any pics taken during the process?

Just one of the GPU taken apart. I can post it if you want.
 
Right now, I'm noticing that the GPU is idling much higher than it was before, so I'm trying to figure that out x.x

Sajin
Increased ambient temps can affect load temps.

I'm running with the AC on, ambient temps should be equal or lower.
I've repasted again (This time with spreading instead of line method)
and my idle temps are still high.

Scarlet-Tech
Please do post the picture.

I do personally spread my TIM when reapplying. For me, it promises to have 100% die coverage where the dot method doesn't guarantee anything at all for coverage of the corners. I tried the xmethod, and actually had massive air bubbles with it twice, so I went back to spreading and have never had an issue. Of course, it comes down to personal preference and experience in the end.

I would love to see your picture you have though.

I attached the photo. This is when I first took the cooler off, the old paste is on. To be fair, looking at it now, it looks kind of like something streaked inside, kind of strange.
 
Anyway, I'm on my 3rd repaste now. I've done it the way EKWB recommends, and I feel like I used far too much paste. I haven't taken it apart, but I'm afraid it might have gotten in the little divet between the die. Any idea how to clean that up?
 
I'll let it sit for a day or so to see if it calms down, but my idle temps are sitting at 60c when they've never been that high with the original paste. Load temps are the same, with similar fan speeds but under idle speeds it's just peaking at the highest idle I've seen.
 
In any case, I'll attempt one more repaste (The line method that worked the best for me) But after that, I'm not sure. The temps are much higher than they were when I bought it, and my case flow is much better than when I first bought the card.

Scarlet-Tech
If the line method worked best for you, stick with it. Make sure your fan header connector isn't slightly disconnected. It may not be fully seated, and giving off inaccurate readings.

I'll double check when I reapply next, but I made especially sure each time I took it apart that the connectors were sound. I've put a lot of time into trying to figure out this temp issue, and to be honest I'd really like to fit it with a hybrid cooler but I've been told the SC+ isn't compatible. :[
 
Oh well, I'll attempt another repaste and if it doesn't iron out my issues I'll just wait until pascal comes out and hopefully attach a hybrid to that when it comes around.

Satzlefraz
I'll double check when I reapply next, but I made especially sure each time I took it apart that the connectors were sound. I've put a lot of time into trying to figure out this temp issue, and to be honest I'd really like to fit it with a hybrid cooler but I've been told the SC+ isn't compatible. :[
 
Oh well, I'll attempt another repaste and if it doesn't iron out my issues I'll just wait until pascal comes out and hopefully attach a hybrid to that when it comes around.

Hopefully the next repaste gets you some better results. Get pictures.. I will always ask for pictures, lol.

Also, I am forgetful and in the process forgot to answer your question about TIM on the capacitors next to the die.

If you are using non conductive TIM, don't worry about cleaning it off. You can't see it with the cover on. This was my latest TIM (I dropped 30c... But I also went from 70c on air cooling to 40c on watercooling, so I cheated basically) :

Which TIM did you use?

Scarlet-Tech

Satzlefraz
I'll double check when I reapply next, but I made especially sure each time I took it apart that the connectors were sound. I've put a lot of time into trying to figure out this temp issue, and to be honest I'd really like to fit it with a hybrid cooler but I've been told the SC+ isn't compatible. :[
 
Oh well, I'll attempt another repaste and if it doesn't iron out my issues I'll just wait until pascal comes out and hopefully attach a hybrid to that when it comes around.

Hopefully the next repaste gets you some better results. Get pictures.. I will always ask for pictures, lol.

Also, I am forgetful and in the process forgot to answer your question about TIM on the capacitors next to the die.

If you are using non conductive TIM, don't worry about cleaning it off. You can't see it with the cover on. This was my latest TIM (I dropped 30c... But I also went from 70c on air cooling to 40c on watercooling, so I cheated basically) :

I always see your posts too late. I didn't take any pictures. D:
I applied the paste yet again, with the line method and see no difference in temps. I cleaned the die pretty thoroughly. I used MX-4 so it's non conductive, there was a little spillage, but I did pretty good at cleaning it.
Someone else mentioned that my PCB/cooler might be warped, and I'm starting to believe them. I was thinking of putting it under water, but I won't if the PCB is warped. I'm not satisfied with these temps, but it's still within spec so there's not much I can do.
 
I appreciate all the help though.
 

Sajin

Satzlefraz

Sajin
Increased ambient temps can affect load temps.

I'm running with the AC on, ambient temps should be equal or lower.
I've repasted again (This time with spreading instead of line method)
and my idle temps are still high.

Maybe the cooler/pcb is starting to warp a bit which is causing the increased temps?

How would I find this out?
It does seem that the heatsink is a little more seperated at the end of the card, but I couldn't remember if it came that way. Pictures look similar to mine, but I'm not sure.

Sajin

Satzlefraz

Sajin

Satzlefraz

Sajin
Increased ambient temps can affect load temps.

I'm running with the AC on, ambient temps should be equal or lower.
I've repasted again (This time with spreading instead of line method)
and my idle temps are still high.

Maybe the cooler/pcb is starting to warp a bit which is causing the increased temps?

How would I find this out?
It does seem that the heatsink is a little more seperated at the end of the card, but I couldn't remember if it came that way. Pictures look similar to mine, but I'm not sure.

Buy another card that is exactly like yours and swap coolers/pcbs and see if your results change.

Hmm, I don't think I would buy another 980ti just to see if my card is defective when I have a three year warranty...

Sajin

Satzlefraz

Sajin

Satzlefraz

Sajin

Satzlefraz

Sajin
Increased ambient temps can affect load temps.

I'm running with the AC on, ambient temps should be equal or lower.
I've repasted again (This time with spreading instead of line method)
and my idle temps are still high.

Maybe the cooler/pcb is starting to warp a bit which is causing the increased temps?

How would I find this out?
It does seem that the heatsink is a little more seperated at the end of the card, but I couldn't remember if it came that way. Pictures look similar to mine, but I'm not sure.

Buy another card that is exactly like yours and swap coolers/pcbs and see if your results change.

Hmm, I don't think I would buy another 980ti just to see if my card is defective when I have a three year warranty...

RMA'ing the card won't let you know if the pcb/cooler is warped.

Wouldn't they replace it if it were?

I've always had best results using the star (or cross) pattern as EKWB recommends. But it really depends on the type of cooler being attached to the card. I found that ASUS non-reference coolers needs a lot of paste to have proper contact. I've seen quite a few positive outcome by using the star pattern, as an example an EVGA 780 Ti Kingpin Edition where there was only dried out paste covering half the GPU, factory installed. That card dropped about 10 degrees Celsius down from 91 by using the star pattern.