Re-Pasted GTX 970 - actually worked....

[Dravinian]

I opened up the Zotac GTX 970, I have two so thought it was worth the risk.

 

Took it apart using a guide for a different GTX 970, but it was useful to watch anyway - doesn't appear to be any guides online for this.

 

The paste on it, well it was barely there, it was dried out completely, patchy and I didn't even need alcohol to take off the paste, I just wiped it off with a cloth.  I did use alcohol to clean it anyway, but really I didn't need to.

 

I used Thermal Gizzly and put a fair amount on the GPU, rebuilt it and put it back in my machine.

Started off on idle at 36c...my second card lower down in SLI was at 29c.  I started up folding to put some load on it.   Ramped up pretty much as normal to be honest, got to about 75c with the fans 100% and at that point I thought, well that hasn't really worked.

 

So I stopped folding.  The GPU went down to 43c and sat there.  My second GPU went down to about 33c and sat there.  The only way I could get the temps down was by making the fans turn on at 20c and attempt to bring it down that way.

 

Do I need to let the paste 'cook' in a bit? Or has it just made no difference?

[Dravinian]

Would the GPU work, if I had not tightened the GPU back to the heat sink hard enough? Could that be a reason that the GPU would still work, and still have the same performance, but simply not be better?

 

I did have them quite tight, but was worried it was too tight, so lossened them and then did them finger tight, in a cross pattern.

[nick name]

Seems like a pressure problem if you only went finger tight.  Try a loose-gripped screw driver tight.  

[Dravinian]

I think I will give it a shot, thanks.

[Dravinian]

It was pretty tight, so didn't tighten it.

Did however decide to do the other one as well, and take the opportunity to put the card that had been on top SLI slot for 6 years in the slave position.

 

Let it bake for a bit, did a Heaven benchmark at 4k, did ramp up.

 

During benchmark, I think I saw maybe a 3c difference.  Which isn't as much as I was hoping for, but is still better than it was before.

 

Then started folding, normally both cards would hit 70c pretty darn quickly, but this time, they took a bit longer to get there, still got there, but it took longer.

 

Gonna let it cook the paste for a bit and see if it gets any better.

 

 

[Dravinian]

I thiink that has done the trick.

 

I wanted to run F@H on this machine, but the fan noise was horrendous. 

 

I realised that in MSI Afterburner, I could turn down the clock and voltage...not just overclock, but underclock.

They were still ramping and the fan noise was still terrible, even on the lowest settings.

After redoing the paste, on the lowest settings, the fans no longer need to kick it.

 

Which means I can leave that one running 24/7 for Folding now....which is nice.

[Blue4130]

33 minutes ago, Dravinian said:

I thiink that has done the trick.

 

I wanted to run F@H on this machine, but the fan noise was horrendous. 

 

I realised that in MSI Afterburner, I could turn down the clock and voltage...not just overclock, but underclock.

They were still ramping and the fan noise was still terrible, even on the lowest settings.

After redoing the paste, on the lowest settings, the fans no longer need to kick it.

 

Which means I can leave that one running 24/7 for Folding now....which is nice.

Changing paste is not going to work magic. A couple of degrees is likely all that you will see (everything else being equal - mounting pressure). Also it sounds like you are comparing gpu in top slot of sli to bottom slot of sli. That doesn't work. One has much more air to cool with. You should compare the same cards before and after repaste. 

[Dravinian]

Just now, Blue4130 said:

Changing paste is not going to work magic. A couple of degrees is likely all that you will see (everything else being equal - mounting pressure). Also it sounds like you are comparing gpu in top slot of sli to bottom slot of sli. That doesn't work. One has much more air to cool with. You should compare the same cards before and after repaste. 

I did actually switch them around as well, to give the top one a break.

 

The top one is always hotter, doing more work, and sitting directly above a hot card.

As I say, it did work, about 3c, not a great deal and I was hoping for more, but it is enough to get the temps low enough that I can fold without fans at 90-100% 24/7