The Official Thermal Interface Material thread

[TPCEA]

I am looking for an end solution for my 7900 XTX air cooler as the hotspot temp has gotten worse. I have already been recommended PTM7950 in the past but I still want to push the card to its limits without it going beyond 95C.

What does this have to do with not choosing PTM7950? It can handle it.

For longevity, Kryosheet, Honeywell, and then Carbonaut.

For reuse and transferability, Kryosheet, and then Carbonaut only.

For electrical insulation, Honeywell only. Those TG products both conduct electricity.

For thermal conductivity, I'm having trouble finding the specs for Kryosheet and Carbonaut, but I saw on TG: "Compared to Carbonaut pads, KryoSheet offers significantly higher thermal conductivity. " As far as I can tell, Honeywell is almost the same as Kryosheet.

For ease of application: Kryosheet, Carbonaut and then PTM7950.

I don't have personal experience with these, so I'm going off data from others and TG. I hope that helps and that someone can elucidate.

 

[freeagent]

I am still using the stock TIM and pads on my 3070Ti and 4070Ti.. no need to change them yet as performance is still excellent. I have the stock pads and SYY on my 980 Classified. Runs pretty good. I also have SYY on my GTX580 Matrix Platinum. I bet you guys would be better off with good case flow to cool the card rather than taking it apart for a few C difference.

 

[wNotyarD]

I am still using the stock TIM and pads on my 3070Ti and 4070Ti.. no need to change them yet as performance is still excellent. I have the stock pads and SYY on my 980 Classified. Runs pretty good. I also have SYY on my GTX580 Matrix Platinum. I bet you guys would be better off with good case flow to cool the card rather than taking it apart for a few C difference.

For me personally, the only way to enhance airflow inside my case is to remove the front panel and I'm not gonna do it.
So I'm willing to repaste the card someday, but I'm in no hurry to do so. Will probably leave it to November when its warranty expires.

 

[physicsjunky]

Performance?

• Best temperatures?
• Longest lasting?

I'm looking for the best temperatures to push the 430-watt 3Ghz OC through the card that I used to get. My GPU is already set up for conductive TIM as I used LM previously.
Now I can't reach 400 watts due to thermal throttling.

 

[Borc]

Dowsil released a new paste called TC-5960: https://www.dow.com/en-us/pdp.dowsil-tc-5960-thermally-conductive-compound.523942z.html#overview

Higher Thermal Conductivity, lower Thermal Resistance and lower Bond Line Thickness than the already great TC-5888. Maybe this is the best paste right now. Not sure how it compares to TC-5550. For heatspreader CPUs the lower bondline from TC-5960 should do better I would assume. Not sure for bare die chips on GPUs or laptop CPUs.

This innovative silicone grease can be used in place of a PCM or thermal pad, and was designed for bare die and lidless CPUs, providing high pump-out resistance and high-throughput.

DOWSIL™ TC-5550 Thermal Conductive Compound

A high thermal conductivity (5.0W/mk), low thermal resistance (0.04 °C-cm2/W), low BLT(0.02mm) thermal grease that is especially designed for bare die architecture to provide long term reliability with excellent pump-out resistance performance.

www.dow.com

It performed well here:

Maybe I can try TC-5550 soon on a laptop CPU which I can compare with TC-5888 and some other best in the market I have in my collection like Shin-Etsu X-23-8117, PTM7950, Thermal Hero Quantum.

 

[Shrek]

I'm glad that pump-out is recognized as a thing.

 

[TPCEA]

Dowsil released a new paste called TC-5960: https://www.dow.com/en-us/pdp.dowsil-tc-5960-thermally-conductive-compound.523942z.html#overview

Higher Thermal Conductivity, lower Thermal Resistance and lower Bond Line Thickness than the already great TC-5888. Maybe this is the best paste right now. Not sure how it compares to TC-5550. For heatspreader CPUs the lower bondline from TC-5960 should do better I would assume. Not sure for bare die chips on GPUs or laptop CPUs.

DOWSIL™ TC-5550 Thermal Conductive Compound

A high thermal conductivity (5.0W/mk), low thermal resistance (0.04 °C-cm2/W), low BLT(0.02mm) thermal grease that is especially designed for bare die architecture to provide long term reliability with excellent pump-out resistance performance.

www.dow.com

It performed well here:

Maybe I can try TC-5550 soon on a laptop CPU which I can compare with TC-5888 and some other best in the market I have in my collection like Shin-Etsu X-23-8117, PTM7950, Thermal Hero Quantum.

What are the top 10 in your inventory?

Performance PCs is having a mother's day sale right now, 15% off everything. If your order costs enough, you can get free shipping. They do sell PTM7950 SP in a 2g tube, which appears it might be cheaper than LTT, especially if there's no shipping.

 

[lexluthermiester]

I'm glad that pump-out is recognized as a thing.

A very limited thing. Not everyone is convinced it is as common as some seem to think..

 

[Divide Overflow]

I'm glad that pump-out is recognized as a thing.

"Pump out" is the new "memory leak" Many claim to have it despite not having the slightest clue what it actually is.

 

[lexluthermiester]

"Pump out" is the new "memory leak" Many claim to have it despite not having the slightest clue what it actually is.

I've never seen any evidence of it IRL. Over that last 4 decades I've handled easily over 30,000 CPU's, GPU's and other chipsets. Problems with TIMs are always instances of not enough applied, applied improperly or the wrong type of TIM was used. "Pump-out" is, to me, a total red-herring(read moose muffins). I'm not denying it happens, but such a situation requires extreme circumstances that are so rare that it is effectively not possible in the consumer or even enterprise sectors.

 

[Borc]

I've never seen any evidence of it IRL. Over that last 4 decades I've handled easily over 30,000 CPU's, GPU's and other chipsets. Problems with TIMs are always instances of not enough applied, applied improperly or the wrong type of TIM was used. "Pump-out" is, to me, a total red-herring(read moose muffins). I'm not denying it happens, but such a situation requires extreme circumstances that are so rare that it is effectively not possible in the consumer or even enterprise sectors.

No wonder when you can't even see MX-4 being poor in many laptop environments.

 

[freeagent]

Quit using that GD900 or whatever it’s called.. boom no more pump out

 

[Shrek]

Anyone here use hard setting TIM?

 

[lexluthermiester]

No wonder when you can't even see MX-4 being poor in many laptop environments.

That's because, and try to follow along here, MX4 isn't poor for laptop environments.

Anyone here use hard setting TIM?

You mean phase-change TIMs?

 

[Shrek]

You mean phase-change TIMs?

No, I mean set dry

 

[lexluthermiester]

No, I mean set dry

In that case, no. Never heard of it. Or maybe I have but am not recognizing your description.

 

[sam_86314]

No, I mean set dry

Like thermal adhesive/epoxy?

Used (way too much of) that stuff to put memory/VRM heasinks on my 1070 when I put an Arctic Accelero on it.

 

[Borc]

THG preview for an upcoming thermal paste review:

https://twitter.com/i/web/status/1790951082648379398

Nice to see that two of the very best laptop pastes tested by me are great on a 14900K desktop chip and even the PTM7950 PCM paste is great there. Remember that Kold-01 has exactly same specs as Dowsil TC-5888, to me it's the same. Dowsil TC-5550 and TC-5960 might do even better.

 

[Shrek]

In that case, no. Never heard of it. Or maybe I have but am not recognizing your description.

When taking apart equipment for the first time I tend to find dryset TIM that can be quite a task to clean off; it seems to be used by the OEM

 

[oiawshboiwsahnoidaws]

THG preview for an upcoming thermal paste review:
Nice to see that two of the very best laptop pastes tested by me are great on a 14900K desktop chip and even the PTM7950 PCM paste is great there. Remember that Kold-01 has exactly same specs as Dowsil TC-5888, to me it's the same. Dowsil TC-5550 and TC-5960 might do even better.

Comprehensive list of the reviewer nice, also cool to see another confirmation that ptm does well on desktop cpus, personally i saw ~2°C improvement compared to kold-01
Interestingly enough the reviewer commented that on watercooling, ptm does not do as well like in his air cooling results

 

[TPCEA]

Comprehensive list of the reviewer nice, also cool to see another confirmation that ptm does well on desktop cpus, personally i saw ~2°C improvement compared to kold-01
Interestingly enough the reviewer commented that on watercooling, ptm does not do as well like in his air cooling results
View attachment 347614

Wouldn't that have more to do with the coldplate than the TIM?

 

[:D:D]

Not sure how it compares to TC-5550.

DOWSIL™ TC-5960 Thermally Conductive Compound

Provides high thermal conductivity, low pump-out and high-temperature stability, designed to provide efficient thermal transfer for the cooling of electronic modules, and good pump-out resistance especially in bare die application.

www.dow.com

Interestingly enough the reviewer commented that on watercooling, ptm does not do as well like in his air cooling results

As phase change temperature can vary between batches SOP after initial installation is to have it at at temperature of 60C for 30mins with a pressure of 30-40PSI. With minimum of 10PSI that becomes 80-90C for 1 hour. Note that at 10PSI expect about 50% more thermal resistance.

Maybe some thermal gels would be interesting?

 

[lexluthermiester]

THG preview for an upcoming thermal paste review: https://twitter.com/i/web/status/1790951082648379398

Yeah, screenshots are more helpful. Twitter/X frequently blocks external links.

When taking apart equipment for the first time I tend to find dryset TIM that can be quite a task to clean off; it seems to be used by the OEM

Ah, I know what you're taking about. That's usually a phase change type that has hardened.