If your GPU has high TEMP!

J

JCNiest5

2[H]4U
Joined
Apr 25, 2005
Messages
4,093
I know this has been posted everywhere on Youtube (and/or everywhere on the internet) already, but just in case you haven't read or watched any yet, here is something you may be missing.

If you have your video card(s) for a few years now and notice that your temp runs higher than in the past and/or weird fans activity, following is what you may want to do:

I have my EVGA GTX 1080 SC for a few years now and just lately, while playing games, the fans tend to be kicking on and off quite abnormally, I took off the heatsink/fan, clean off the old thermal paste really nice and apply brand new one, it really helps. Now, the fans don't go crazy anymore.

Before: Temp during games, 89-91

Now: Hovering 77-79

Just thought I'd share this with you computer maniacs out there.
 
FrgMstr said:
Maintenance. Whodathunkit?
Click to expand...
That's some really shitty paste choice if a relatively new 1080 requires a new application. That's not maintenance, that's a piece of shit product.
 
Meeho said:
That's some really shitty paste choice if a relatively new 1080 requires a new application. That's not maintenance, that's a piece of shit product.
Click to expand...

Cheapest route possible, the way forward......forever. This is nothing new, shit paste has been applied to any consumer grade thermal component for decades.
 
FrgMstr said:
Three years old is relatively new? Those GPUs get really hot....
Click to expand...
So do Intel CPUs and those run far longer. I would say three years is quite new for a PC component regarding its lifespan.

Other GPUs from other manufacturers, even other EVGA cards, run for longer without the paste running out and causing a >10 C delta.
 
Meeho said:
So do Intel CPUs and those run far longer. I would say three years is quite new for a PC component regarding its lifespan.

Other GPUs from other manufacturers, even other EVGA cards, run for longer without the paste running out and causing a >10 C delta.
Click to expand...
This has nothing to do with components, this has to do with the TIM. There are many different variables that can affect those over time. An application that was not quite right, install that was not quite right, movement of the heatsink over time, all can introduce oxidizers into the TIM causing it to become somewhat compromised over time. So yes, if you are seeing inflated temps over time with similar ambient temperatures, a TIM reapplication is likely a good place to start if it has been years.
 
I can see how this can easily be overlooked. I have never removed the heatsink to apply TIM of any of my graphics cards over the years. A great reminder!
 
always good to replace TIM. every 3 years isn't bad practice, and if you don't normally blow the dust out of your case, do that at the same time.
 
deaedius said:
I can see how this can easily be overlooked. I have never removed the heatsink to apply TIM of any of my graphics cards over the years. A great reminder!
Click to expand...
I probably could have gotten a few more years out of my 8800GTX pair had I not ignored it.
 
I tend do reapply thermal compound at least yearly, if not more often but that's because I enjoy doing it and if I happen to have the PC open for some other reason, it won't hurt to do it while I'm in there. I also blow dust out of my PC, clean the air intake filters, and other menial PC tasks.
 
Hmmm... I understand why, but I haven't reapplied tim to anything unless I had to disassemble it for cleaning or upgrade. I build my own PC's and typically apply some Arctic silver on assembly and remove whatever crap comes from the factory. Haven't had issues in many years through many PC's.
 
I've never once had to reapply TIM. I've blown plenty of dust out of HSF units over the years since my Voodoo 4500 AGP and weird Pentium 4 on a Pentium 3 socket, but never actually had to replace TIM.
 
How often should someone reapply TIM?

I've never reapplied the TIM on any of my GPUs (my oldest one still in active gaming service is a reference GTX 780 while the very old ones just normally have desktop duty). The original owner of my Palit 1070 opened up the heatsink probably to replace the TIM, but managed to damage the fans and the shroud, so I'm also a little wary about disassembling GPUs...

My CPUs get a fresh paste of TIM whenever I remove the cooler, although this doesn't happen often, and sometimes I get lazy and put the cooler back on anyway.
 
mda said:
How often should someone reapply TIM?

I've never reapplied the TIM on any of my GPUs (my oldest one still in active gaming service is a reference GTX 780 while the very old ones just normally have desktop duty). The original owner of my Palit 1070 opened up the heatsink probably to replace the TIM, but managed to damage the fans and the shroud, so I'm also a little wary about disassembling GPUs...

My CPUs get a fresh paste of TIM whenever I remove the cooler, although this doesn't happen often, and sometimes I get lazy and put the cooler back on anyway.
Click to expand...

I'd say no point changing it if you haven't had the same CPU or GPU for over 3 years. If you remove the cooler then new TIM should be applied every time.

The thing is that on GPUs the stock TIM is usually applied very generously at the factory and usually replacing it can drop temps up to 5 degrees, allowing for possibly lower noise or higher boost clocks.
 
Armenius said:
I probably could have gotten a few more years out of my 8800GTX pair had I not ignored it.
Click to expand...

I would definitely have gotten another few years out of my XFX 8800GTX if hadn't overclocked it to around 640Mhz on the core and left it running SETI all day in summer. Ouch.
 
mda said:
How often should someone reapply TIM?.
Click to expand...

There's no set rule, because all TIM are different and all use cases are different. Basically, just change it when you notice temp issues. Most TIM should be good for 2+ years. In my experience, many have a hard time going past 5 years.
 
mda said:
How often should someone reapply TIM?

I've never reapplied the TIM on any of my GPUs (my oldest one still in active gaming service is a reference GTX 780 while the very old ones just normally have desktop duty). The original owner of my Palit 1070 opened up the heatsink probably to replace the TIM, but managed to damage the fans and the shroud, so I'm also a little wary about disassembling GPUs...

My CPUs get a fresh paste of TIM whenever I remove the cooler, although this doesn't happen often, and sometimes I get lazy and put the cooler back on anyway.
Click to expand...
Most compounds start to dry out at around the 3 year point, but it varies by manufacturer since everyone has their own proprietary formula.
chameleoneel said:
There's no set rule, because all TIM are different and all use cases are different. Basically, just change it when you notice temp issues. Most TIM should be good for 2+ years. In my experience, many have a hard time going past 5 years.
Click to expand...
Was going to say. I would say just do it every 3 years, or every 6th time you open your PC up to clean it biannually like I do (dust filters, etc.). In my experience AS5 lasts 2-3 years, MX-4 could probably go 5 years, and Kryonaut should go no longer than 2 years. I don't have experience with NT-H2, but Noctua says it lasts 5 years when applied to a CPU.
 
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
Armenius said:
Most compounds start to dry out at around the 3 year point, but it varies by manufacturer since everyone has their own proprietary formula.

Was going to say. I would say just do it every 3 years, or every 6th time you open your PC up to clean it biannually like I do (dust filters, etc.). In my experience AS5 lasts 2-3 years, MX-4 could probably go 5 years, and Kryonaut should go no longer than 2 years. I don't have experience with NT-H2, but Noctua says it lasts 5 years when applied to a CPU.
Click to expand...

Do the compounds have to be 'wet' to function optimally? I do notice that the compounds eventually dry out, but then again I have a few Pentium 4's doing relatively well being used in an office environment wherein we've never had to part the heatsink and the CPU.
 
mda said:
Do the compounds have to be 'wet' to function optimally? I do notice that the compounds eventually dry out, but then again I have a few Pentium 4's doing relatively well being used in an office environment wherein we've never had to part the heatsink and the CPU.
Click to expand...

I don't think they do. I've replaced some TIM on mobo northbridges that are so damn dry it takes some work to get it off. When I replace it with quality TIM the temps improved but it was nothing amazing. I'm sure there's a point where it is necessary and dependent on what kind of TIM they used but lots of times it still works dry.
 
mda said:
Do the compounds have to be 'wet' to function optimally? I do notice that the compounds eventually dry out, but then again I have a few Pentium 4's doing relatively well being used in an office environment wherein we've never had to part the heatsink and the CPU.
Click to expand...
There is a curing point where the compound is adhered to the mated surfaces and then there is a "drying out" point where the compound is no longer active and not transferring heat optimally. The situation described above is the latter. When a compound dries out it will be like concrete and you'll have to chip away at it, and when you do it will crumble. A cured paste will feel dry to the touch, but it will still have a luster to it when you look at it and it will have more of a peeling action if you try to take it off without using alcohol.

It is typically only a few degrees difference because the point of TIM is to fill in the imperfections of the two mated surfaces to create an even surface for contact to occur. A "dry" TIM will act as an insulator of heat instead of a conductor.
 
I've taken off the heatsink of a 2600k after 7-8 years in service and paste looked just as fresh as day 1. Of course paste brands may vary in quality and some will dry out quicker than others. But never considered anything under 3 years use as requiring new paste (as long as its decent paste).
 
I was watching a GN video interview of der8auer from thermal grizzly. Basically, the newer pastes on the market that are easier to spread and have very good initial thermals also suffer from a phenomenon he called "pump out" Basically the heating/cooling cycles in a cpu/gpu die create a pumping action that eventually pushed the paste away from the center of the IHS or die and that is why you see degraded performance after a couple of years or less.

He also said that OEMs usually go for a paste that has less thermal performance but MUCH higher resistance to "pump out"
 
Armenius said:
There is a curing point where the compound is adhered to the mated surfaces and then there is a "drying out" point where the compound is no longer active and not transferring heat optimally. The situation described above is the latter. When a compound dries out it will be like concrete and you'll have to chip away at it, and when you do it will crumble. A cured paste will feel dry to the touch, but it will still have a luster to it when you look at it and it will have more of a peeling action if you try to take it off without using alcohol.

It is typically only a few degrees difference because the point of TIM is to fill in the imperfections of the two mated surfaces to create an even surface for contact to occur. A "dry" TIM will act as an insulator of heat instead of a conductor.
Click to expand...

Der8auer of Kryonaut says that "curing" and "drying" don't actually exist. According to him the only time drying is an issue is if you're using very cheap paste with a low quality oil base.
 
Last edited:
You must log in or register to reply here.
Back
Top