GPU applied Kyronaut now 3C hotter..

[Upgrayedd]

I replaced the TIM on my EVGA GTX 970 SC and the stock paste looked pretty dry. I was sitting in the PUBG idle screen getting up to 73C with the old stock TIM lately. I thought it was probably time to change it and I might see a small drop. My optimistic guess was going to be 69C but after I applied the new TIM, Thermal Grizzly Kyronaut, my PUGB idle screen temp is back to 73-75C. I never seen this card hit 75C. Anyone else tried this on their GPUs? I had a tough time spreading it but it spreads great on the gritty CPU lid.

 

[Solaris17]

I did mine fine. I would try again and be double sure the block is seating correctly.

 

[jboydgolfer]

A common misperception is that if a thermal paste isn't expensive, it isn't effective . However, human error is often the culprit in situations like this more often than not, & is also the case in this specific situation of yours in my opinion.

 

[erocker]

Maybe there's too much gap between the GPU and the cooler?

 

[Upgrayedd]

I used 70% Isoproply but I am pretty much out. I am going to go get some 98% right now and try a reapply. TBH it doesn't really feel like the GPU heatsink screws down very tight, not near as tight as my Noctua at least.

 

[jboydgolfer]

I used 70% Isoproply but I am pretty much out. I am going to go get some 98% right now and try a reapply. TBH it doesn't really feel like the GPU heatsink screws down very tight, not near as tight as my Noctua at least.

Although cleaning old paste off is important ,the highest purity of isopropyl alcohol is not required. Just use some tissue paper ,and wipe it until it stops making the tissue dirty, water will help if you run out of alcohol , alcohol is best though, but only a few drops are required, & only after youve removed all but residue.

The simplest solution would be to clear a space on a flat clean service, entirely disassemble the card, remove the old thermal paste with a tissue or paper towel, once the majority is removed and when all that's left is a little film ,use a couple drops of isopropyl alcohol , and clean the remaining TIM off. Then apply a small drop of new thermal paste directly in the center of the GPU chip, then take your time and remount the shroud and heat sink, I'm sure it will work out fine. Be sure to pay close attention to any thermal pads that maybe on memory chips etc. verify that they're making proper contact & not coming loose or have any debris on them.

 

[cadaveca]

Iso isn't going to remove any wax build-up.

 

[johnspack]

God please don't use 70%, that's 30% water! Get iso 99%, and get a bottle of acetone. Clean the surfaces with the acetone and cuetips, then use the iso 99 to finish.
The iso also denatures the acetone residue, and removes it....

 

[jboydgolfer]

God please don't use 70%, that's 30% water! Get iso 99%, and get a bottle of acetone. Clean the surfaces with the acetone and cuetips, then use the iso 99 to finish.
The iso also denatures the acetone residue, and removes it....

iirc, Acetone can damage PCBs , he'll be entirely fine using any percentage alcohol, I've been doing it for decades.

The only place any alcohol or any type of solvent should be going is on the end of a balld up piece of tissue paper ,and then rubbed to remove residue or film build up that may be remaining after he wipes the GPU and the heat sink clean , so as long as he's careful I'm sure acetone would be fine as well, but i want to atleast mention, in case my recollection is correct regarding Acetone and PCB's.

 

[eidairaman1]

iirc, Acetone can damage PCBs , he'll be entirely fine using any percentage alcohol, I've been doing it for decades.

The only place any alcohol or any type of solvent should be going is on the end of a balld up piece of tissue paper ,and then rubbed to remove residue or film build up that may be remaining after he wipes the GPU and the heat sink clean , so as long as he's careful I'm sure acetone would be fine as well, but i want to atleast mention, in case my recollection is correct regarding Acetone and PCB's.

Been using 70-90% Isopropyl Alcohol, even Hydrogen Peroxide, no problems

 

[Upgrayedd]

This how she looked after attempt 1 and a few C hotter. I just tried my 2nd application and now it was going up to 79C so I just exited the game.. I feel like I am doing something wrong..In the picture it shows on the corners where there is no contact. I am tightening the screws in a cross pattern.

EDIT: So after that really bad 79C 2nd attempt and noticing it wasn't getting completely even pressure even though the screws were all the way in and cross tightened, I went a 3rd go and pushed on the back of the card around the die as I tightened the screws and also after I was done I just gave it a few more pushes on the back to try and make the surfaces meet. On the 3rd go she is finally sitting in the main menu of PUBG at 69C. I also decided to keep the surface of the heatsink clean when I mounted the two instead of putting TIM on both sides like a lot of people suggest, just seems like too much material to me but if you're having mounting contact issues then I could see it being a bit useful.

EDIT 2: Ok having a bigger problem now. I am getting constant crashes during games, one told me I ran out of video memory, others were just "application stopped working". Once I crash to desktop my background is black and can be revealed by using the blue box used to select multiple desktop icons. Sometimes the background just decides to go to black. Really hope I didn't fudge something.

EDIT 3: It is the next day now and the night before I noticed my pc only showed 8GB of ram when I have 16GB, I guess I didn't push the stick all the way back in. Pulled the stick out and made sure I heard it click this time. Showed 16GB now but was still getting crashes left and right in games. Used DDU, restart and reinstalled the driver I was always using before. It was still crashing after a full shutdown for 20min and then after a another restart..
I was getting frustrated, I decided to restart again and check the card in the BIOS and it showed it running full x16 lanes, so I just used boot-override to boot to the OS drive because I didn't change anything and everything works fine now. I was crashing ten seconds into RotTR benchmark consistently but now I can run it and my games just fine. Don't know what happened really.

 

[Vayra86]

Stop fiddling with these things because DRY PASTE is HOW IT GETS and is NOT AN ISSUE. So you gained 4 C... nearly killing some hardware in the process.

Damn... people these days.. This procedure should go in the 'Don't flash GPU Bios' category.

 

[Upgrayedd]

I hate to say this but I already predicted a dead card in the opening post and its starting to look an awful lot like it.

Stop fiddling with these things because DRY PASTE is HOW IT GETS and is NOT AN ISSUE.

Damn... people these days.. This procedure should go in the 'Don't flash GPU Bios' category

I actually have flashed the GPU and the card is fine now. Just didn't smash the RAM stick all the way in I guess. Been playing games, no crashes and no artifacts..How do you figure I had a dead card? from my temps? Why do people change their TIM every few months?

 

[Vayra86]

I actually have flashed the GPU and the card is fine now. Just didn't smash the RAM stick all the way in I guess. Been playing games, no crashes and no artifacts..How do you figure I had a dead card? from my temps? Why do people change their TIM every few months?

I misread - I thought your crashes weren't fixed Edited later.

Regardless the point stands. People who change their TIM on regular basis have lost the plot. Its not needed 'because its dried out' and every time you fiddle with this stuff you introduce risk to the component(s) through human error, ESD, etc etc and the benefits are minimal.

 

[jboydgolfer]

I actually have flashed the GPU and the card is fine now. Just didn't smash the RAM stick all the way in I guess. Been playing games, no crashes and no artifacts..How do you figure I had a dead card? from my temps? Why do people change their TIM every few months?

I don't know anybody who changes the thermal paste on their devices every few months. If there is somebody who does that ,they need to get a life or find something to do in their free time ,because it's pointless and senseless and silly. Here I was thinking changing thermal paste frequently was doing it every year or two . I think the biggest mistake people make, is they buy a device and immediately feel that they should replace the thermal compound, like they're so good, & all knowing ,that they're stupid little tube of thermal paste is going to "fix" mistakes of a multibillion dollar company , please. The Arrogance is strong

@op.
As far as your situation ,just leave it alone ,it'll be fine

 

[BarbaricSoul]

You are using too much TIM. You should only be using a pea sized dab of TIM. Do not apply it to both the GPU and heatsink. Too much paste causes temperature to run higher than they should. Don't spread it out either. Just put a pea sized dab of TIM on the GPU and install the heat sink. Let the heat sink being squeezed against the GPU spread the TIM.

 

[John Naylor]

1. I use Indigo extreme cleaner but in a pinch will use 97% alcohol which last I bought was $2.59 at Walmart. Remeber gpoing from 97% to 70% you are increasing the amount of impurities left behind by a factor of 10 ! Technique is important.... and if you are only applying TIM to the GPU, you are doing it wrong. I use Shin Etsu G751 ($4) on CPUs but like many other TIMs it's not well suited to GFX cards cause it can take time to apply to memory and VRMs which can mean application to 20 or more surfaces. TIMs like Gelid Extreme are recommended as they have longer "workability" times. Gelid also supplies an application tool which is recommended .... you will notice that the "pea" method is NOT recommended for GPUs ... see installation guide for a 970 below.

https://www.ekwb.com/shop/EK-IM/EK-IM-3831109869406.pdf

2. Recognize that the 970 SC is just a nVidia refence card with a better cooler and has known issues one of which the lack of thermal pads on PCB....same ones which caused the recall / 'thermal kit" thing on the 1000 series 1060 - 1080. Refer again to above PDF

https://www.bit-tech.net/reviews/tech/graphics/nvidia-geforce-gtx-970-review/3/

a) if you have one of the early ones, a design error resulted in 1 of the 3 heat pipes missing the GPU ... oops
b) "The GPU MOSFETs are directly cooled by the main heatsink, which has a thermal strip on to draw heat up into the fin stack. On the other side of the GPU is a metal contact plate that partially cools two of the four memory chips on this side, leaving the other two exposed. It also cools the MOSFETs of the power phases serving the memory, but no thermal pads are used, so heat transfer is likely to be limited"

So when applying paste, remember that you need to make sure the thermal strips are replaced and its advisable to address the missing ones if possible....make sure all contact points get TIM replaced.

 

[Upgrayedd]

Yeah I thought people were over the top on changing TIM so often but I only decided to do it because it had been a few years and seemed like it was getting a bit hotter.
@John Naylor,

this what I'm dealing with, uneven mounting pressure or not enough pressure, I used criss-cross screw tightening method multiple times. I finally just pushed on the back of the PCB as I tightened it down and seems to have brought temps back to normal but nothing actually lowered, just has fresh TIM I guess lol.

 

[phanbuey]

Some tims work better on cpus and some on gpus - but a 1-2c difference is really margin of error. The temp in your room might be hotter etc

 

[Arrakis9]

Yeah I thought people were over the top on changing TIM so often but I only decided to do it because it had been a few years and seemed like it was getting a bit hotter.
@John Naylor, View attachment 101476this what I'm dealing with, uneven mounting pressure or not enough pressure, I used criss-cross screw tightening method multiple times. I finally just pushed on the back of the PCB as I tightened it down and seems to have brought temps back to normal but nothing actually lowered, just has fresh TIM I guess lol.

Make sure all of the hex nuts on the cooler it self are tightened down all the way. I've had it happen more than once were a nut un tightens as you pull a screw out.

 

[Vario]

I have found that when you repaste graphics cards you have to make sure the paste is well spread across the whole die, use a baggy on your finger. It is also different from CPUs in that you can and should use more thermal paste then the grain of rice. That is not to say use a ton of paste, but I found better results using more than on a CPU. The excess just smooshes out, good reason to use a non conductive paste or just be strategic in the amount you use with a baggy. Sometimes you will get a result that delivers good temperatures but still artifacts when under load, this happens if the thermal paste is not evenly spread on the surface of the die. You really want a smooth uniform grey layer of paste over the entire die. It is similar to doing a laptop as well.

When you tighten down you have to be careful to make sure it is even pressure and uniform pressure, as you did. I also found that some heatsinks have screws that are narrower diameter than the holes in the PCB which can cause the heatsink to shift and the thermal paste to unseat, so I added a bit of heat shrink tubing to the screws on my heatsink to remove the slop. This worked really well.

 

[hat]

For GPUs at least I've had better luck squeezing out an appropriately sized blob/line of paste and letting the heatsink squish it down, rather than spreading it myself. For HDT coolers like in post 18, I squeeze a really thin line on each heatpipe, then let the cooler squish that.

 

[bubbleawsome]

It looks like the corner isn't making contact at all. Is it catching on the side of the die or something?

 

[Upgrayedd]

It looks like the corner isn't making contact at all. Is it catching on the side of the die or something?

I'd have to look at it again, I tried looking in there to see if the board seemed bent from sagging, it did seem a little flexed. I might take it out and try it again, it isn't too hard or time consuming. You can see that the surface contact is worse in post 11 than in post 18, maybe it just purely my fault, I will have to check the nuts on the cooler like Arrakis+9 suggested, really didn't feel like much pressure to be honest.

EDIT: In the past I have had great luck on my CPU applications where I put the thinnest layer I can, it is basically a waxy haze. I am going to try this as I felt everytime I was putting it on the GPU I was doing too much, it just wasn't spreading good for me so I put more than I normally would.

For GPUs at least I've had better luck squeezing out an appropriately sized blob/line of paste and letting the heatsink squish it down, rather than spreading it myself. For HDT coolers like in post 18, I squeeze a really thin line on each heatpipe, then let the cooler squish that.

Going to try that if other attempt doesn't show anything.

 

[Space Lynx]

For GPUs at least I've had better luck squeezing out an appropriately sized blob/line of paste and letting the heatsink squish it down, rather than spreading it myself. For HDT coolers like in post 18, I squeeze a really thin line on each heatpipe, then let the cooler squish that.

I do this on CPU's and GPU's with my Noctua NT-H1 paste, just a big blob of it right in middle, my temps always seem slightly better than everyone elses average by about 1-3 celsius. ::high five m8::