6970 lightning high temp, some sort of residue on the core? (pics)

Hello all,

I've been having high temp issues with my MSI GPU, it was mostly the 3rd reported GPU temp in GPU-Z, it would reach 100 to 107 degrees under 100% load, 1st temp ranged between 55-65 only and the 2nd temps would get in the 70s range. I've had this issue before and reapplying the thermal paste made it better (3rd went down to mid 80s), cleaning the dust off didn't solve the issue.

So I removed the heatsink to reapply the thermal paste (I use Noctua NT-H1 non-conductive paste that came with my CPU cooler), cleaned off the old paste with some iso propyl, but there was some very light residue that I couldn't clean off on one part of the core, some pics (sorry for the bad quality, used a crappy phone camera, hopefully the negative image will show it better):
http://img844.imageshack.us/img844/841/dsc01667jj.jpg
http://img21.imageshack.us/img21/8329/dsc01674yk.jpg
http://img856.imageshack.us/img856/9067/dsc01678p.jpg

It's only noticeable when you put some direct light on it. After reapplying the thermal paste (I put a small blob and spread it using my finger wrapped in plastic bag), temps are even worse now, even the 2nd GPU temp would reach 100+ along with 3rd, 1st temp is even cooler than before though lol

So I guess I fucked up reapplying the paste using the spread method, so before I reapply it Again I thought I'd ask a couple of questions:

1- What do you guys think is that residue thing? It looks like dried out water or something, I've used iso propyl 70% now and before to remove the old paste, could that be it? i.e it didn't dry fast enough? Could this be the reason why only some temps of the GPU die is getting hot while the 1st temp remains cool (3rd temp specially always ran hot compared to the other 2)?

2- It's not getting removed which is pissing me off, any ideas about how to do that safely if possible? This whole exposed core thing makes me pretty paranoid to try anything on it, wish AMD didn't use such design

3- Shall I try another paste? I know people mostly say that it's all about the application, but I don't know, the noctua paste is so damn thick and pretty annoying to apply, plus I'm still a beginner at this

If you're wondering why I didn't just RMA the card, well it's because I ordered it from the states and shipping would be pretty expensive, so I'm keeping that as the absolute last resort (AFAIK reapplying the thermal paste doesn't void the warranty, at least there weren't any seals to break when removing the heatsink).

Any help would be really appreciated.

Looks like some sort of composite build up, probably from the heat drying up the paste. Worrisome that it's not coming off.

If RMA is your absolute last resort, you can try another paste or application methods although I doubt they'll make a difference.

I had a 6870x2 and I had a bad application on one of the cores as I was reseating the heatsink and had temps like yours but you say you always had high temps. I'd go straight for the RMA and save the headache.

Thanks for replying.

I really would prefer to keep the RMA as a last resort, gotta try something first because shipping will probably cost me around $150-170 which is quite a lot. I just have a feeling (wishful maybe?) that this residue is causing some parts of the core to get hot, any ideas about methods to dissolve such stubborn stuff safely would be great.

I was thinking about getting Arctic silver thermal removal and surface purifier, been also reading about using charcoal lighter fluid too, don't know if that's safe on an exposed core, sounds kinda risky.

That's a crazy amount for shipping!

The bottles of arctic silver removal and purifier work great, i'd try it and maybe even goo gone.

Never heard of charcoal lighter fluid used to remove residue but it's possible since oil removes certain materials. Can try vegetable or olive oil too if you're going all out lol.

sai_dee said:

That's a crazy amount for shipping!

Click to expand...

Yeah international shipping costs a fortune, the price is cheaper using regular postal mail services instead of a courier, but that would take forever to deliver and there's no tracking so it's too risky.

Goo Gone looks interesting, but it seems like it's exclusively sold in the states, I'll try to see if I can find any. There's no way I'd try out vegetable oil though, that sounds pretty crazy even to me lol

I asked arctic and they confirmed that using their cleaner is safe on an exposed core, I'll try to get some tomorrow if possible and try it out, will post my results when I do so.

Try some denatured alcohol or maybe acetone - you can find these things at a hardware store. Acetone will destroy plastics pretty fast, so avoid using it around the shroud, fans, etc.

Looks like it over heated and the material started to melt. I'd RMA it. Looks damaged.

Only store that had arctic clean is out of stock now, this sucks I was really looking forward to try it out.

sonsonate said:

Try some denatured alcohol or maybe acetone - you can find these things at a hardware store. Acetone will destroy plastics pretty fast, so avoid using it around the shroud, fans, etc.

Click to expand...

Any specific type of denatured alcohol or acetone I should look for, like pure or specific %..etc? I know I heard the nail polish remover acetone has some oil in it and shouldn't be used or something

Since the core seems to have a metal surface, using those materials should be fine as long as I make sure I'm not going to spill anything over the rest of the components?

SonDa5 said:

Looks like it over heated and the material started to melt. I'd RMA it. Looks damaged.

Click to expand...

I gotta try something first, $150-170 for international shipping is almost the price of a new mid range GPU lol

So options that I'm thinking of using so far are:
- Acetone
- Denatured alcohol
- Try to look for something that's called "Goof Off", it's even mentioned on Arctic Silver's website as a method to remove old thermal pastes since it's xylene based or whatever the heck it's called lol

The acetone in the hardware store shouldn't have any additives, but it might leave a residue since it's an oil product, but since it's a solvent it might remove the residue you have. Whatever little left over residue, after it evaporates, could be wiped off with some isopropyl.

As for the denatured alcohol, I use this one: http://www.amazon.com/Klean-Strip-Denatured-Alcohol-Gallon/dp/B001EMB2A0

I couldn't tell you what other specifics, since I never tried any other one. Cleans without melting things, lol.

Goof-off will eat plastic as it is a pretty much like acetone - a solvent.

Thanks again for replying.

I couldn't find pure acetone or denatured alcohol, but I did get my hands on a bottle of Goo Gone and another bottle of Goof Off, I'm going to use one of them tomorrow then clean whatever is left with iso propyl and hope for the best.

I heard using both is a bad idea (at the same time anyway), so if anyone has any experience with either of those products, please let me know so I'd have a better idea about which to use tomorrow.

While I've never used either product, I'd use them 1 at a time, applying them to the rag and wiping the core, followed by an isopropyl wash. BUT, looking at your pics, it looks like the core isn't flat and is cooking the paste where there is a lack of contact pressure. It looks as though the die is too low, or the shim or it's bonding agent is too thick. You can check this with a new razer blade, and using it as a straight edge. Rest it across the shim, sharp edge straight down, and see if the core is touching the blade while it rests on the shim. Try it in several different directions. I've seen it in the past with the x800 series cards, and would explain why you can't correct it with TIM replacement. At that point you need to either remove that perimeter shim and lap it down (shim) so the core can contact the heatsink, or put a thin copper sheet from a machine shop as a shim between the core and the sink so they make contact. If you choose to remove the shim, be careful, as if you bend it, it's nearly impossible to flatten again. I can't guarentee this is your problem, but if I had to bet on it, that's the way I'd go

Just wanted to update you guys about what happened in the past few days:

- Used goof off and it didn't remove the residue, cleaned with iso propyl then used blob in the middle method (my NT-H1 paste finished with this application). The results were absolutely terrible, both my 2nd and 3rd GPU temps would hit 100+, my 3rd GPU specially went over 110 degrees, I didn't even wait to see how far it would go because I feared I'll damage the card permanently.

- Had to look around for some good non-conductive paste and found Antec formula 7 , it's supposed to be diamond dust based or something and non-conductive.

-Opened up the GPU Again (I swear I could do this blindfolded by now lol), removed the old paste with alcohol and then tried to remove the residue stuff with Goo Gone and it also didn't work, cleaned whatever is remaining with iso propyl and made sure the surface is as clean as possible, put a nice line of Antec formula 7 paste, instead of putting the heatsink on the core, this time I put the PCB on the heatsink so I wouldn't have to turn the thing around to screw it.

Results:
IT WORKED! I don't know if it's because the paste is so good (doubt it) or because I used a different application method and connected the PCB to the heatsink in a different way (that would be my guess). Max temps I'm getting now is high 50s/low 60s for the first temp and 70s for temp 2 and temp3 when running 100% load (heaven benchmark, MSI kombustor..etc) and the fan is under 50% which is pretty damn quiet! Playing Skyrim used to easily push the 3rd temp to 100+, now it's mostly in the low 70s (around 30 degrees drop!)! MSI tech support said that the max safe temp for my GPU core is 102, so there's a lot of headroom left even if I wanted to overclock now (I won't though lol).

It's been a couple of days and I stressed the GPU enough by now, so the minimal burn-in time should be up and the temps will most probably stay like this =) Idle temps are also great at 35-38 degrees only across all 3 temps.

I have a feeling I didn't put enough paste before (after hearing lots of people suggest to NOT use a lot of paste, specially for a small GPU core surface), it seems to me people exaggerated about the whole use as little paste as possible thing, all I know is that I'm a believer in putting a nice line method now and going to use the Antec F7 paste since it worked for me. Never going to use the spread method ever again.


tl;dr: Goof-Off/Goo Gone didn't work, alcohol did the same job and it's easier to clean. Spread and blob methods using NT-H1 failed leading to horrible temps. Used Antec Formula 7 paste with a nice line application method, temps drops around 30 degrees and everything is great now (max 70s temps).


@FragMagnet
I didn't have a razer to test that, I couldn't detect any issues with how flat is the core when looking at it though, maybe my crappy pictures weren't clear enough. Hopefully I won't have to reopen the GPU yet again for a long time to do that test lol