***GTX980 G1 teardown, clean and reassembly with Thermal Grizzly Hydronaut and results***

[stulid]

Here is a little guide to show you how to take apart a GTX980 G1 gaming card from Gigabyte, clean the GPU core surface and re-apply fresh thermal paste before re-assembly.


Items Needed

• One GTX980 G1 gaming GFX card (Windforce models will be similar, minus the backplate).
• A small, cross-head screw driver.
• Thermal paste remover or IPA and/or surface purifier.
• Toilet roll and cotton buds.
• Thermal paste.


Disassembly

Here is the card I will be using. I have already given the backplate a slight clean as it has been used since October last year and was dusty.

This is the screw driver I will use and it's the only tool needed for this job.

Take these three screws from out of the rear - at these three highlighted holes.

Loosen these four screws and remove them. (I have already started this which is why they look a bit loose already in this photo).

Using a bit of force (the two halves have been together since purchase and the existing thermal paste may have made a tight bond) the cooler will come away from the PCB+Backplate. Then unplug the two cables (fans+illuminated logo).

You should now be at this stage:

To take the backplate off unscrew these six screws.

Now you should be looking at this:


Cleaning

These are the consumables I will be using:

I started by wiping the core with some toilet roll in order to get the bulk of the paste off. Then, using a cotton bud, take the paste around the edges off. After applying the thermal paste remover and a few more cotton buds and toilet roll I got it looking like this.

Going gently, I cleaned all traces of old paste off the little SMD parts around the core.

As this was solid copper I went a bit harder on it and used thermal paste cleaner until the toilet roll I was using stopped looking dirty. I then used the surface purifier on both parts.

So you will now be at this stage.


Removing all the Built-Up Dust

If your card has been in use for awhile it may have built-up some dust deposits - mine certainly had. If yours is cleaner, this next step is optional.

Using a hairy brush attachment on a vacuum cleaner, I removed the dust.

I also cleaned this slightly dusty area up.

That is about it for the PCB; the back can be given a clean too considering the backplate was taken off earlier.

I also wanted to take the cooler apart further to clean all the fins and fan blades.

Undo the four screws highlighted above and the fan and shroud assembly will come away from the heatsink.

This is the main dust build-up from months of use.

I went at these two parts again with the hairy brush attachment on my vacuum cleaner, paying extra attention to the fan blades and getting deep into the heatsink fins.

It now looks like this: almost box-fresh again.


Reassembly

Putting the card back together is just a reverse of the above.

I re-attached the backplate to the PCB using the six screws and re-attached the fan and shroud assembly to the heatsink.


Thermal Paste Application

To mount the heatsink back onto the PCB I used Thermal Grizzly Hydronaut paste. This paste is formulated for use with water-cooling, but hey...let's see what it can do in this application.

The supplied tube of paste has a very efficient applicator that makes spreading the paste an absolute breeze.

A nice coverage of paste.

So all that is left to do is place the PCB onto the upside down heatsink and screw those four screws with tension springs back in - again, a few turns at a time and in opposite corners, and the three final screws at the end of the card.


The Reassembled Card

Looks as good as the day I bought it.


Results

Before disassembly, I ran GPU-Z render test for 30mins and two runs of Unigen Valley. I then repeated this after using the new paste.

Core temperature was recorded by GPU-Z logging to file - I used the peak value reached during this and deducted the ambient temperature.

I also tested using auto-fan speed, setting the fan control to a fixed 40% in MSI Afterburner.

I then repeated these four tests with the new paste.

Ambient temperature was recorded using my digital Hilka thermometer and the corrected Delta temperature is shown below.

So that has knocked between 1.3°C and 1.7°C from my cards' temperatures; every little helps I guess.

 

[pastymuncher]

It will probably make a bigger difference on the cpu. For some reason I never see as big a drop on the gpu as I do on the cpu. Isn't that layer of paste you applied a bit thick? I usually have a layer at least half as thick as that. I wouldn't have used a hoover to clean the card. Knowing my luck the static would end up killing the card. I find a 1/2" paint brush does the job well enough.

 

[Dicehunter]

Although the process you went through is appreciated the temp difference hardly seems worth it TBH.

 

[stulid]

It will probably make a bigger difference on the cpu. For some reason I never see as big a drop on the gpu as I do on the cpu. Isn't that layer of paste you applied a bit thick? I usually have a layer at least half as thick as that. I wouldn't have used a hoover to clean the card. Knowing my luck the static would end up killing the card. I find a 1/2" paint brush does the job well enough.

Obviously I do have CPU testing to come.

The layer of paste is fine and will spread out more once the cooler is clamped back down.

I have never had a problem using a hoover that is earthed to the mains.

Although the process you went through is appreciated the temp difference hardly seems worth it TBH.

Considering its managed to reduce the temps on already the best cooler ever fitted to a GTX980, I think its done ok.

 

[Razor Time]

IMO you applied way too much paste. You might improve your temps more with less TIM.

 

[stulid]

IMO you applied way too much paste. You might improve your temps more with less TIM.

It is a very thin layer, trust me on this, it just covers so damn well and isnt transparent like other pastes or leaves gaps as you spread it.

 

[Yaayuh!]

Nice write up, but the difference is so minimal it could be put down top dust and change in ambient.

 

[stulid]

Nice write up, but the difference is so minimal it could be put down top dust and change in ambient.

I have measured the ambient and the temps given are corrected for this.

 

[varkanoid]

Is it stuff that improves over time ?

 

[stulid]

Is it stuff that improves over time ?

No, non curing.

 

[Razor Time]

It is a very thin layer, trust me on this, it just covers so damn well and isnt transparent like other pastes or leaves gaps as you spread it.

I'm just going off your pic with the paste spread across the chip. It really should be a semi transparent layer.

I wonder if you would also get a bit more off with using gelid thermal paste. I've not used the paste you have tried so unfamiliar with its performance.

 

[eyetrip]

Unless it improves over some time, I would say it's not worth it for 99.9% of owners to go to the trouble.

But at least people can see now what they should be getting, and what they may gain, so this will be a handy reference for many G1 owners.

 

[stulid]

I'm just going off your pic with the paste spread across the chip. It really should be a semi transparent layer.

I wonder if you would also get a bit more off with using gelid thermal paste. I've not used the paste you have tried so unfamiliar with its performance.

You cant get it transparent as that is not what this paste looks like when applied thinly. I have used many pastes and done many applications, some look like the above and some look see through

Its nothing like this cheaper paste,

And here is another paste thin layer that isnt transparent

So as you can see it depends on the paste as to whether it should be semi transparent or not

 

[andybird123]

I'm just going off your pic with the paste spread across the chip. It really should be a semi transparent layer.

I wonder if you would also get a bit more off with using gelid thermal paste. I've not used the paste you have tried so unfamiliar with its performance.

it's very similar or arguably slightly better than the gelid (as in fractions of a degree better)

 

[stulid]

Guys!!

Remember this is the Hydronaut paste not the Kryonaut paste that matches/beats the Gelid according to 8pack

 

[Razor Time]

Sorry mate I dont know why there was an angry face in my reply I was posting from my phone.

If your happy with it the results thats all the matters. It does show that gigabyte hadnt done that bad of a job in pasting the block originally. I've seen some blocks with enough TIM to do 3 cards and a good 5c+ drop when repasted.



AndyBird - Ah a new alternative. Nice one as I mentioned I hadn't heard of it before.

 

[colinmcr]

What a great guide and write up thank Stulid

 

[HEADRAT]

Agreed, its very good guide for those wanting to take the plunge (I have voted this thread five stars)

 

[Dirk Diggler]

I go for a nice blob of paste in the middle, I hate spreading it. Paste of choice for me is Arctic Silver.

To be honest, the guide is handy to see how the card comes apart but I wouldn't be taking it apart unless I was putting a water block on. I also think that the choice of paste isn't so important, and going by your temperature "drop" it looks like the case here.

A slightly more aggressive fan curve would have given better results?

 

[bru]

Nicely done stulid, thanks for taking the time to write that up with some nice clear pic's.