A thermal phase change material is a possibly. It's solid when cold like wax. It can fill a small gap. I've seen this on various GPUs. ATI / AMD in particular.

You will need a shim or pad as a replacement. Paste can not fill a gap. I'm not sure a pad has enough thermal conductivity to be used with a high end GPU. It might be possible to find a patch of phase change stuff too. It used to be sold in squares, for AMD Athlon CPUs because AMD recommended it.

Thanks for the info/suggestions. So phase change material would be waxy, not chalky, is that correct? The stuff I just removed was chalky so I assumed it was dried out paste from a previous cleaning.

I found a picture of what is presumably an unused cooler from a 280:
http://www.ebay.com/itm/Lot-of-2-Cooler-Maste … =item4190f369f5

From appearance, it looks like it has the same chalky gray stuff that I just removed from my card, except mine wasn't in a neat square like that, it was a circle that didn't quite cover the whole chip. At this point I'm not sure if the stuff I removed was original phase change material, or dried out paste that I had put there. I thought I had disassembled this heatsink before, but I might be mixing it up with another card.

I think my plan will be to try to reduce the clearance and see if I can get the parts to touch, so regular thermal paste can be used. With luck, maybe the RAM and GPU will have similar clearances. If not, well at least they'll be closer together. If there continues to be a gap at the RAM I'll just replace the thermal pads on those. If there's a remaining gap at the GPU, maybe I'll try a thermal pad or I'll look into getting that phase change stuff.

One other problem with my 260 that I didn't get into is that I made a mistake taking it apart and broke most of the screw posts on the plastic shroud. Therefore I decided to put it back together without the plastic and screwed a fan directly into the heatsink. The fan has no shortage of power but the cooling is lackluster, I think because of the GPU-heatsink gap already discussed. I probed the heatsink with an infrared meter and could hardly find much heat on it except in one spot underneath the fan blades.

I made the mistake of looking at eBay prices for similar cards (initially I was just interested in a shroud donor) and ended up buying a working GTX275 which appeared to be properly tested at a good price. Of course, it may have just as bad or worse cooling issues as my 260, especially since the coolers are apparently the same.

That ebay photo looks like paste to me. I remember NV usually has lots of that grey paste slathered on their GPUs. Maybe they do fill the gap with it...

This comes to mind. It's pretty obvious that the paste wasn't pushed out much and so there probably is a gap.
http://ixbtlabs.com/articles3/video/gt200-par … 0-part1-p6.html

Here is a Radeon 6970 with phase change material. It's fairly hard and crusty when it's cold.
http://www.ixbt.com/video3/cayman-part2.shtml

fill the gap with a piece of copper shim. It will work much better than filling the gap with paste. Even an alu shim is better.

I saw various cards with thermal pads on the gpu from both camps. these gaps are a big reason why 3th party coolers often work way better, they simply fit better and are made with much smaller tolerances.

I removed the pads once on my x1950xtx to replace them with good paste. It was a dissaster and i was glad i didn't throw away the pads and removed them in one piece. They were back on really quick.

RAM pads are fine. But I didn't realize that some of these cards had such poor GPU contact.

Too much thermal paste = bad.
Using an copper film (thin plate) with paste will do wonders as suggested. There are also 4W/km pads.

My GTX295 had some kind of ceramic cloth pad.

Stojke wrote:

Too much thermal paste = bad.

Apparently it worked acceptably for millions of NVIDIA's cards though. The paste used may be engineered for this purpose. Who knows. It's certainly not optimal however.

My Cousin had given me his old PNY 260(216) but it was overheating and locking up, this is why:

You can see the Thermal pads had degraded and fallen apart

Nasty

After everything was cleaned and ready to be reassembled

I didn't take a photo of it reassembled, but there was a small gap on the ram modules at the top of the card. However it ran just fine without lockups. Gave it to my dad and it's still in use today.

ZanQuance wrote:

My Cousin had given me his old PNY 260(216) but it was overheating and locking up, this is why:

Sounds like the same card I have. The gray stuff on the GPU in your picture looks like what was on mine. However, mine had long strips of black thermal tape over the RAM (covering multiple chips at once), not individual white pieces like yours.
I noticed on eBay that there seem to be 2 different versions of the cooler on the PNY GTX260-216 cards. Not sure there's any real difference between them beyond the plastic shroud though.

The problem I think I'm having is with those gray screw posts visible in your 3rd picture. I think they're too long, at least on my card, causing the heatsink to have too much clearance. Whenever I get back to working on this card I'm going to see if I can shorten those. Putting a copper shim on the GPU would probably work just as well - I just don't have one and I think it might need to be epoxied permanently to prevent any chance of it sliding out.

The BFG GTX275 I bought on eBay showed up, and it cools much better than my PNY 260-216. I bought my 260 when new and I'm not sure it ever cooled this well. I'm surprised - I feared since they appear to use the same cooler, the 275 might be worse. I was also worried about my PSU - it's shown signs of being on the edge with the 260, but it's had no problems so far with the 275 so I guess I'll just keep using it.
As much as I'd like to see what the heatsink clearances are like on the 275, I don't want to take it apart.

=====
Something I just learned that I want to share. I never realized it before, but by default, these GT200 cards will constantly run at full 3D clocks if you have dual monitors attached. However, there's a workaround if you use the program called "NVidia Inspector". In that program, you can right click on the "show overclocking" button and there will be an option for a "multi-monitor power saver". This will force the card into 2D clock rates. You can set up rules for when it will switch back into 3D clocks. The only wrinkle I've seen is that when the clock rates are changed, the screen will blink momentarily. Not a big deal as long as your configuration doesn't have it constantly switching back and forth.

Old thread I know
But I believe I have found the root cause of the overheating GTX2xx cards.
The problem lies with the PNY/EVGA magnesium mold casting of these coolers.
The mounting posts are not level or entirely too tall.
In my case with a EVGA GTX260-216 I had absolutely horrific temperatures going above 110C and the fan going into warp 9.
Then I tested another reference GTX260 and it also had terrible temperatures, but not over 100C
Main thing that caused alarm was the rate of temperature rise, several degrees per second as if there was no thermal mass attached to the IHS.

I then tested the gap using a feeler gauge and well there was no gap, there was a divide several continents wide between the GPU and Heatsink.
I ended up filing down the mounting posts almost 2mm and the heatsink - GPU contact has improved tremendously.
The cards now run at 60C under full load with 40% fan speed.

Too much thermal paste = bad.

Common myth. It's doesn't matter.

This is how bad the thermal interface is by default.
No pressure applied to the blobs of paste at all.
Not something that can be realistically solved by just adding more paste.

A 9500GT recently came to me that was running abnormally hot. The paste is shot I thought. No big deal. I took off the HSF and...

there was a piece of paper between the GPU and the thermal pad.

Clearly these pads are shipped to the factory on strips of paper which you are supposed to peel off, and someone forgot to do this. (Or is this automated nowadays?)

Removing the old paper and pad, and applying thermal paste, caused the idle temperature to drop 10 degrees C.

Pads are not applied as GPU thermal interface for a long time now. So it was most likely aftermarket solution.

0xCats wrote on 2020-10-27, 09:59:

Old thread I know But I believe I have found the root cause of the overheating GTX2xx cards. The problem lies with the PNY/EVGA […]
Show full quote

Old thread I know
But I believe I have found the root cause of the overheating GTX2xx cards.
The problem lies with the PNY/EVGA magnesium mold casting of these coolers.
The mounting posts are not level or entirely too tall.
In my case with a EVGA GTX260-216 I had absolutely horrific temperatures going above 110C and the fan going into warp 9.
Then I tested another reference GTX260 and it also had terrible temperatures, but not over 100C
Main thing that caused alarm was the rate of temperature rise, several degrees per second as if there was no thermal mass attached to the IHS.

I then tested the gap using a feeler gauge and well there was no gap, there was a divide several continents wide between the GPU and Heatsink.
I ended up filing down the mounting posts almost 2mm and the heatsink - GPU contact has improved tremendously.
The cards now run at 60C under full load with 40% fan speed.

Hello. I am "restoring" right now an EVGA gtx260.I have completely dissasembled the card, cleaned it, but i just can't figure out what thermal pads i need and how much thermal compound is needed on the GPU. I would like to know what thermal pads did you use and what solution have you found to the heatsink-GPU contact problem, to make it run so cool. Thanks