Should I repaste my MSI 1080Ti Gaming X?

Nebell

Nebell

2[H]4U
Joined
Jul 20, 2015
Messages
2,383
I'm going to sell my two 980Ti's and I need to remove waterblock from them and mount original coolers.
I bought Arctic Cooling MX-4 and read it's a good paste. Since I'm already fixing my 980Ti's, should I repaste my 1080Ti as well or just leave it as is? Will there be any difference?
 
I had 2 MSI 1080s and I did a repaste, they did have a high amount of paste gobbed on from the factory. I will say it looked like high quality paste though, probably arctic silver. It did lower temps by a few degrees.
 
Sprayingmango said:
I had 2 MSI 1080s and I did a repaste, they did have a high amount of paste gobbed on from the factory. I will say it looked like high quality paste though, probably arctic silver. It did lower temps by a few degrees.
Click to expand...
The amount of paste is totally inconsequential. Every GPU manufacturer uses enough to spooge over the side of the GPU die. Far better than having any part of the bare silicon not covered.

That said, my 1070 has Thermal Grizzly Kryonaut on it and it helped a bit, but it really doesn't matter how much there is as long as the entire piece of silicon is covered, and too much is better than not enough.
 
I mean, unless the thing is thermal throttling constantly or just hitting unreasonable temps I don't know what the point is tbf.
 
GlacierNine said:
The amount of paste is totally inconsequential. Every GPU manufacturer uses enough to spooge over the side of the GPU die. Far better than having any part of the bare silicon not covered.

That said, my 1070 has Thermal Grizzly Kryonaut on it and it helped a bit, but it really doesn't matter how much there is as long as the entire piece of silicon is covered, and too much is better than not enough.
Click to expand...


Interesting. I feel like someone tested this before and found that too much is bad....not enough is bad....and somewhere in the middle is optimal. I did see a reduction after doing the repaste
 
I dunno who you saw test it, but here's LinusTechTips video on paste application for CPUs -

Also, JayzTwoCents did this video on the topic:



Both of them conclude that too much thermal paste is a total non-issue, and will only cause problems if the thermal compound ends up getting into some component or other that it shouldn't. If it's just a bit of extra mess, then you have nothing to worry about.

Bear in mind also, that a lot of the received wisdom about thermal paste is old as hell, and things have changed in the meantime. Mounting systems in particular, for coolers, are far better engineered and put on a lot more even, consistent pressure. Cooler bases are machined flatter, and the IHS on a modern CPU is pretty flat too, in stark contrast to the old days where waterblock makers would deliberately make their blocks convex, to account for Intel's poor IHS flatness.

Hell, how many 2002-2004 era heatsink reviews have you ever read where paste thickness was mentioned? If you've been screwing around with PCs as long as I have, probably a bunch, but those guys were testing Socket 754/939 and Socket 478/LGA775 coolers that attached with clips and pushpins, not the high-pressure, high rigidity backplates we're all used to since the mid 00's brought tower coolers to the masses and demanded stronger mounting methods.

I suspect that these days, any good cooler will display zero difference in performance with any pasting method, UNLESS there is simply not enough paste to cover the whole die/heatspreader. As a result, rather too much than not enough.
 
Last edited:
This is also why I remain an advocate of the oldschool "spread the goop across the die/heatspreader with a tool of some kind" method of thermal paste application. (Back in the day this would be a credit card, these days companies like Thermal Grizzly and Gelid provide helpful silicon spatulas)


It ensures complete coverage of the heatspreader or die. The rice grain methods or line methods always leave a circular blob of goop on a square IHS, and while not a problem on an IHS where the die, and therefore center of heat output, is only under that blob anyway, on a bare-die solution like a GPU, it's very important to ensure there's paste covering every available bit of silicon.

The fear with the credit card method was always that you'd trap an air bubble. I think it's pretty clear these days that air bubbles simply don't get trapped like that. and I just remembered another excellent video about thermal compound application methods, that pretty much proves it beyond a doubt -



If anyone is going to know their shit, it's a man whose job is to OC with LN2.

It shows the "paste" style compounds essentially just spread however is needed.

Liquid metals tend to shoot out of the sides, and thus need spreading in advance to avoid getting the stuff everywhere where it shouldn't end up. Other than that, go nuts.
 
Well, I did repaste my MSI 1080Ti Gaming X and it made no difference. Maybe even increased the temperature by 1 celsius. Now it's at 74-75C in Unigine (maxed at 4k) while boosting to 1980MHz. I used Arctic Cooling MX-4.
Before I did this, I disassembled my MSI 980Ti Gaming 6G and repasted and it's reporting 78C in Unigine, while Inno3D 980Ti Ultra X3 is only at about 70C.
 
GlacierNine said:
This is also why I remain an advocate of the oldschool "spread the goop across the die/heatspreader with a tool of some kind" method of thermal paste application. (Back in the day this would be a credit card, these days companies like Thermal Grizzly and Gelid provide helpful silicon spatulas)


It ensures complete coverage of the heatspreader or die. The rice grain methods or line methods always leave a circular blob of goop on a square IHS, and while not a problem on an IHS where the die, and therefore center of heat output, is only under that blob anyway, on a bare-die solution like a GPU, it's very important to ensure there's paste covering every available bit of silicon.

The fear with the credit card method was always that you'd trap an air bubble. I think it's pretty clear these days that air bubbles simply don't get trapped like that. and I just remembered another excellent video about thermal compound application methods, that pretty much proves it beyond a doubt -



If anyone is going to know their shit, it's a man whose job is to OC with LN2.

It shows the "paste" style compounds essentially just spread however is needed.

Liquid metals tend to shoot out of the sides, and thus need spreading in advance to avoid getting the stuff everywhere where it shouldn't end up. Other than that, go nuts.
Click to expand...


THIS!!!!!

I just don't get it when people still advocate so vociferously for one method or another. Especially as I assume they're not rigorously testing back to back for themselves.
 
  • Like
Reactions: Caedo
like this
I wouldn't bother. You'll gain, at best, -2 to -3oC from your max temp. That's not gonna be much of a difference.

Unless you're removing the heatsink to put in a different one/WC block there's no point.
 
Fan curve is more of a issue than the Thermal paste I found out. When the card is on Auto instead of Agressive or a custom curve it will heat up to 84.
So just get rid of the Auto cooling and you should be fine. The program doesn't work very well Precision XOC doesn't give you any prompts that the fan curve is saved and off Auto cooling.

I just fire up Superposition Benchmark and look at the results of the Fan curve to see if it can stay under 70C anyway.
 
I swapped out TIMs on my 1070 and it knocked off a few degrees. Nothing extravagant tho.
 
I just changed paste on my 1080 TI FE when I put my EVGA hybrid cooler on so to be honest with you, there was a huge difference but 100% due to the cooler and not the paste on the other hand I just recently picked up a GTX 1060 Mini as my new backup and I did change the paste on it only because it was so easy to take the cooler off and it did shave 4C for Max load Temps so nothing huge but noteworthy.
 
Just make sure you're using a non-conductive and non-capacitive compound if you're going to have spillover, which the MX-4 is. Anything with silver in it, like Arctic Silver 5, is going to be capacitive.
 
Jayham said:
I just changed paste on my 1080 TI FE when I put my EVGA hybrid cooler on so to be honest with you, there was a huge difference but 100% due to the cooler and not the paste on the other hand I just recently picked up a GTX 1060 Mini as my new backup and I did change the paste on it only because it was so easy to take the cooler off and it did shave 4C for Max load Temps so nothing huge but noteworthy.
Click to expand...

I will be doing exactly this upgrade in the next week or so. Part of me wants to repaste with the stock cooler and measure like for like, but ultimately the hybrid cooler is a temporary solution until Fall when my Calyos chassis arrives. Not thinking I'll do too much over locking on this card, did that with the 680s and these days I care more about component longevity.
 
I would, on one of my Powercolor 7970;s I was getting some high-ish temps when brand new, I was also fighting with a hot IOH on my x58 board. When I re-did the TIM on the board I figured fuck it and did the 7970 as well.

Replaced with some cheap Xigmatek (still decent tho) and dropped load temps by 15*. Worth the 10 minutes and 4 screws if you ask me
 
You must log in or register to reply here.
Back
Top