Question GTX 1070 8gb what Thermal pads do i need?

[Sova001]

I got a 1070 8GB and i used 1mm Thermal Pads on it...how ever it made my temp's high as in 80-90*C

i looked online and noticed some sources say for ones near the chip set for GPU u need 0.5 and for the for the last 2 u need one 1mm and one 2mm...please let me know if this is all correct? i ordered or that is i had a friend get for me a 0.5 hoping that is all i need to fix the issues cuz the 1mm are to thicc so it dosent let the heat sink connect to the chips and paste? or am i understanding it wrong?

 

[Phaaze88]

Changing thermal pads can be an expensive trial and error process. It is not enough to know the depth of the original pads, because pads also come in different degrees of hardness, leading to scenarios where:
A)Gpu core thermals suffer because the replacement pads were too firm; contact between the cooler cold plate and gpu die is now distorted. [Your current scenario.]
B)Memory thermals suffer because the replacement pads were too soft; contact with the cooler heatsink was poor. [GTX 10 series gpus don't have Vram monitoring available without the use of external devices, so if you did this wrong, you wouldn't even know it...]


You need to hear from someone who has had success replacing pads on that exact gpu, and the exact brand and model of pads used. I am not that person. I know the above because of reading multiple, unsuccessful attempts from others.
I changed the cooler on my 1080Ti Gaming OC to a Kraken G12 back in June 2020. I didn't bother with thermal pads on the memory, because from reading around, I found that GDDR5X memory does not have as high operating thermals of R6 and R6X and it can be cooled on good airflow alone. 2 years later, it's still chugging along. No artifacts and whatnot that appear from overheating Vram - but I also don't bother with Vram OCing.


Maybe consider doing away with pads, and being sure the card is getting adequate airflow over it.
Good luck.

 

[Sova001]

Changing thermal pads can be an expensive trial and error process. It is not enough to know the depth of the original pads, because pads also come in different degrees of hardness, leading to scenarios where:
A)Gpu core thermals suffer because the replacement pads were too firm; contact between the cooler cold plate and gpu die is now distorted. [Your current scenario.]
B)Memory thermals suffer because the replacement pads were too soft; contact with the cooler heatsink was poor. [GTX 10 series gpus don't have Vram monitoring available without the use of external devices, so if you did this wrong, you wouldn't even know it...]


You need to hear from someone who has had success replacing pads on that exact gpu, and the exact brand and model of pads used. I am not that person. I know the above because of reading multiple, unsuccessful attempts from others.
I changed the cooler on my 1080Ti Gaming OC to a Kraken G12 back in June 2020. I didn't bother with thermal pads on the memory, because from reading around, I found that GDDR5X memory does not have as high operating thermals of R6 and R6X and it can be cooled on good airflow alone. 2 years later, it's still chugging along. No artifacts and whatnot that appear from overheating Vram - but I also don't bother with Vram OCing.


Maybe consider doing away with pads, and being sure the card is getting adequate airflow over it.
Good luck.

Is it okay if i remove the pad around the GPU core to see if temps improve? the pads before were almost gone as u can see in the image the 0.5mm ones (current ones on them are the 1mm that i mentioned witch i think are the issue :/ )

 

[Phaaze88]

Is it okay if i remove the pad around the GPU core to see if temps improve? the pads before were almost gone as u can see in the image the 0.5mm ones (current ones on them are the 1mm that i mentioned witch i think are the issue :/ )

If they're the original pads, I don't see any problem with leaving them there.

 

[Sova001]

If they're the original pads, I don't see any problem with leaving them there.

they where removed as i stated the 0.5mm ones the old ones like in the image where replaced 2 days ago with 1mm :/
I ordered the 0.5mm ones cuz it seems that the 1mm around the GPU core are the issue only :/

 

[InvalidError]

The simplest way of accurately measuring mounted gap widths would be to put small modelling clay balls on everything, mount the HSF, remove it and measure the squished balls and pick one size up.

Me, I'd try to find steel, aluminum or copper scraps of near enough thickness to fill the remainder with thermal paste.