Nvidia GPU Thermal Paste Replacement

[Grilled Salmon]

So, I have an EVGA 1070 8 GB graphics card, and I was wondering if it'd be safe to take the GPU apart and replace it's thermal paste? And would anyone have any form of link to a video I can watch as a tutorial, since I obviously don't want to destroy my 700 dollar computer part?

P.S. Sorry for second post, I figured this had nothing to do with the PhysX problem, so I decided to make a separate post.

Edit: https://valid.x86.fr/kbc0lh My CPU-Z Verification. Just a heads up, the Motherboard has been changed since this one, but I figure that can't matter that much when it comes to GPU thermal paste.

[Illusion of Progress]

For most GPUs, you'll have four (maybe more) screws on the backside around the GPU chip itself which holds the heatink to the card, and then extras for the shroud or other heatsinks (VRAM and/or VRMs, if any). I'd just search Google or Youtube for your particular model.

I recently did this with my EVGA GTX 1060 and surprisingly, found little in the way of my particular model (most were for the smaller form factor one). I also ended up tearing one of the wires out since it was so hard to remove, but I knew this was just for the white LED feature (I have no other lights in my PC so I was okay using this). FWIW I saw no temperature drops although the paste was slightly (not overly) hard. At this age it's worth doing for these cards IMO.

[_I_]

most will use thermal pads on mosfets and vram
you can replace the thermal paste on the gpu core

most are fairly easy, remove a few screws and carefully unclip the fans from the heatsink and remove the heatsink

half pea to grain of rice sized dot is all you will need

[[☥] - CJ -]

i redid the paste on my 1060 a while back, dropped temps up to 30C without an overclock, with a high overclock temps dropped 10-15C or so depending on the game.

So yes, it is a good idea to redo the paste now and then because factory paste jobs tend to suck or use cheap paste.

I would take pics of which screws go where though, and be careful with the fan header.
-

Add some paste to the GPU and evenly spread it around with an old credit card or something so its fully covered (dont add an excessive amount, just enough to over the GPU when spread), i also add a small dab in the middle so it spreads when the cooler makes contact.

(I say this because while the CENTER of the GPU is important, the corners also carries heat, so theres no reason to not make sure the entire GPU has coverage.) Not once has this method failed me, many others do it that way as well.

Just make sure to have some isopropyl alcohol to clean off the old paste from the GPU and the Cooler first.
-

Plenty of youtube videos to look up, from reputable sources.

[🆂🆃🅸🅶🅶🅰]

引用自 _I_：

most will use thermal pads on mosfets and vram
you can replace the thermal paste on the gpu core

most are fairly easy, remove a few screws and carefully unclip the fans from the heatsink and remove the heatsink

half pea to grain of rice sized dot is all you will need

Yep put to much on and it`s a bad as none, with a CPU to much sticks the heat sink fan to the CPU chip set , need a crow bar to get it off

[Set-115689]

Are the temperatures ok? If they are I would leave it alone. You can seemingly do every thing right and end up with a gpu that doesn't power on.

Don't use an electrically conductive paste. You might also have to replace some thermal pads if they get destroyed. Thermal pads are not all one thickness. If you end up with sections of the die not having paste you might get weird unstable gpu behavior. Will probably also need a high percent isopropyl alcohol to clean off the paste. Don't knock off the little parts/components around the gpu die. Work in a non carpeted area and don't wear staticky clothes like a sweater. Remember to unplug the wire when you separate parts or you'll rip it out. There might be tear down videos for your exact card on youtube. Do it when your comfortable and in an ok frame of mind and have lots of extra time. And again if the temperatures are ok I would just leave it be. If cards get more available and cheaper then that will be in your favour if things go wrong.

[Illusion of Progress]

引用自 Set-115689：

Are the temperatures ok? If they are I would leave it alone. You can seemingly do every thing right and end up with a gpu that doesn't power on.

You might also have to replace some thermal pads if they get destroyed. Thermal pads are not all one thickness.

This is why I hesitated to do mine until I did. I only ended up doing it since I was doing as full PC cleaning. All I've ever done with that GPU since 2016 was give it compressed air treatment two, maybe three times, since 2016, so I expected to be surprised with dust on the fans, within the heatsink, etc., but there was almost none. My temperatures were fine (a bit warm, but always were) and with the GPU shortage, the last thing I needed was to break something... which I did anyway, but luckily it was inconsequential.

I'm not sure about OP's GTX 1070 but if it's the ACX3.0 model like my GTX 1060 is, and if they are the same (for the GTX 1060 and 1070), then there should be no additional heatsinks for the VRAM or VRMs, nor no thermal pads anywhere to worry about (but OP should overlook the card to see if theirs has any to be sure). On mine it was just one heatsink that affixes to the GPU itself, and a shroud over that, with two fans on the heatsink. But there's also like at least 3 different AC3.0 models for the GTX 1060 and I have the lesser one, so I'm not sure if the SSC and FTW are just higher boosts alone (I "think" they are) or also have more physical features which the OP's GTX 1070 may have.