[SOLVED] Liquid Metal

[stanley88845]

I have some liquid metal left over from my cpu installation. I was thinking of putting it on my gpu as the temps sit around 65°-70° which is stopping me from doing any significant oc. If I apply it would I have to tape off the pcb surrounding the dye? And this sounds so stupid but electrical tape is safe for that right?

 

[Barty1884]

It would be best practice to protect around the GPU die, yes.
Electrical tape is 'safe', yes. Certain nail polishes could be an option too.

IIRC, liquid metal on a GPU doesn't yield gains proportionate to the risk. You might shave off 1-2'C, nothing significant.

The rationale being liquid metal between die & IHS on a CPU is one thing - there's an additional barrier. You're already making direct die to cooler contact on a GPU. A quality thermal paste should yield similar results, without the inherent risks (however minimal the risk might be)

 

[hotaru.hino]

Are you sure the temperature is holding the OC back? The temperature limit on most modern cards tends to be 80C by default and if you're using a tool to overclock, you can raise that limit. So I don't think it's temperature holding the OC back. Run some benchmarks with GPU-z in the background show the Sensors tab and monitor the "PerfCap Reason." Unless it's "Thrm", it's not GPU temperature related.

 

[stanley88845]

Are you sure the temperature is holding the OC back? The temperature limit on most modern cards tends to be 80C by default and if you're using a tool to overclock, you can raise that limit. So I don't think it's temperature holding the OC back. Run some benchmarks with GPU-z in the background show the Sensors tab and monitor the "PerfCap Reason." Unless it's "Thrm", it's not GPU temperature related.

Definitely is the temperature, not because it hit its limit but because of the rule that lower temperatures mean great stability. At 100% fan speed I can run a plus 50mhz oc on memory and core with core voltage maxed out just fine. But on a normal fan curve it crashes with any overclock. Temperature limits don't mean much when ocing as the higher the temperature the more unstable it is, it'll probably crash before hitting temp limits. My gpu has a super high boost clock stock anyway at around 1950mhz-2000mhz when gaming and is stable. With and oc I can get it to 2000mhz-2100mhz but after around 10-20 minutes it hits around 70° and crashes. if I could just get 5° less then I could run it at that oc, and a 5° difference should be the minimum I get with my card.

 

[Barty1884]

and a 5° difference should be the minimum I get with my card.

Between the stock TIM and LM, that's likely.

If you're already replaced the stock TIM with something halfway decent, don't get your hopes up for a 5'C reduction. Not saying it's not possible, just unlikely.

 

[stanley88845]

Between the stock TIM and LM, that's likely.

If you're already replaced the stock TIM with something halfway decent, don't get your hopes up for a 5'C reduction. Not saying it's not possible, just unlikely.

Considering the fact I have a gtx 1070 that ends up being on 100% usage at least a couple hours a day everyday for quite some time, I'm pretty sure the thermal paste is going to be dried up as I've never replaced it. And going from dried up thermal paste to liquid metal will be a pretty decent jump.

 

[stanley88845]

Just replaced the thermal paste. Temps dropped from 73° with no oc down to 61° with a +25 core voltage, +75 core clock and +100 memory, could definitely go higher with the clocks.