Any one tried graphite thermal pads?
Was thinking of getting a few of those for rack mount server applications.
Echo what elmariachi said on TFX. I've opened my card up more than three times. Tried MX-4, NT-H1 and temps increased after a 2-3 months each time, with widening hotspot deltas (15c increased to about 20c).
Switched to TFX - initial results were the best of the lot, and there's still no significant deterioration until now (I think it's been about 6 months). Only 0.1c increase in hotspot delta after 6 months, which is very impressive.
Kyronaut is known to breakdown past 80, but in this case, it is more for where the contact die isn't perfectly flat, like the 3000 series gpus. Breakdown =! pump out.okay, thanks, will see how first.
I guess all these scenarios are more likely to occur on components that produce a lot of heat. Such as RTX 3070 and above.
For peasant rigs that produce less heat, I don't think will breakdown or dry so badly? Cos apparently kryonaut still sitting at number 1 for fresh application..
So is it only for high-end parts that produce more heat is affected.. or all ways affected?
Kyronaut is known to breakdown past 80, but in this case, it is more for where the contact die isn't perfectly flat, like the 3000 series gpus. Breakdown =! pump out.
Pastes like TFX work better in such a use case (and also for laptops, where the mounting pressure is crap, because it is thick enough that it doesn't pump out over time.
I've used the same 3 pastes on my 5950x, which is a much flatter surface. They all perform similarly, with no noticeable deterioration over time, and the 5950x generates a lot more heat than my 3080.
Nope. Most 3080s will start throttling at 80+ for core, and 110 for memory. But mine never crossed 75 core and 84 memory undervolted, and the deltas still deteriorated over time.oh.. doesn't the 3080 generate more heat? Plus applying to direct die and not IHS. I would think that it's more likely to breakdown in 3080.
Nope. Most 3080s will start throttling at 80+ for core, and 110 for memory. But mine never crossed 75 core and 84 memory undervolted, and the deltas still deteriorated over time.
My 5950x hits 80 regularly and goes to 90 on prolonged workloads.
oh okay, i thought people would increase the limit to 93 or something coz nvidia claim no problem.Nope. Most 3080s will start throttling at 80+ for core, and 110 for memory. But mine never crossed 75 core and 84 memory undervolted, and the deltas still deteriorated over time.
My 5950x hits 80 regularly and goes to 90 on prolonged workloads.
I normally will take the thermal conductivity rating with a pinch of salt. If it is a non-conductive paste, there is a limit as to how much heat it will transfer. So in real life, the drop in temp is also going to be limited. And I also feel that the problem with most thermal compound review is that they don’t do a long term test. So like Thermal Grizzly’s paste, they work really well at the start, but start drying up after a few months of usage.https://www.igorslab.de/en/alphacoo...-conducting-paste-from-testing-into-practice/
This paste with impressive Thermal conductivity 17 W/mk
Chipset cooling usually takes a back seat. So you can expect the manufacturers to use average quality thermal compound or pad, unless they specifically mentioned that better solutions are used. So no surprises when you replace it with a good thermal compound.After more than 2 years, decided to apply TFX on the x570 chipset(the fan is always off). Temp drops from low 60°C to 50°C. Didn't expect it to drop 10°. Hence can confirm that the fanless x570 chipset design is BS.
Chipset cooling usually takes a back seat. So you can expect the manufacturers to use average quality thermal compound or pad, unless they specifically mentioned that better solutions are used. So no surprises when you replace it with a good thermal compound.
I normally will take the thermal conductivity rating with a pinch of salt. If it is a non-conductive paste, there is a limit as to how much heat it will transfer. So in real life, the drop in temp is also going to be limited. And I also feel that the problem with most thermal compound review is that they don’t do a long term test. So like Thermal Grizzly’s paste, they work really well at the start, but start drying up after a few months of usage.
From what I observed, NVME drive will increase the chipset temp. The rest of the components are not so much.The quality of the chipset cooling also depends on the components we used and stuff into the system usb/pcie connector/slot.
So far as I am aware southbridge chipset degradation/issue less prone compare to system northbridge.
I wonder how many really suspect their system issue was due to southbridge chipset instead of cpu/ram/gpu.