thermal paste Which Thermal Paste to buy and apply (Traditional and Liquid Metal)

[mido]

3 hours ago, Falkentyne said:

 

TFX keeps performing decently over time but the temps DO degrade.  It's a paste, not magic, and no paste is immune to degradation.  Seen it on my RTX 3090 FE.  Lost about 5C over 4 months.

The best way to avoid degradation is to use liquid metal (usually stable long term, after losing a couple C after a godly temp reduction on direct die (BGA CPU's, desktop GPU's, while it's stable on top of IHS and under IHS due to increased mounting pressure), or to use phase change products, which are largely immune to any of these effects while not being as effective as overall LM, it's more stable long term.

 

https://www.aliexpress.com/item/3256803559684957.html?gatewayAdapt=4itemAdapt

 

https://www.moddiy.com/products/Laird-TPCM-780SP-High-Performance-CPU-Interface-Material-(W{47}mK->-5.5).html

How well do the films work? I remember similar solutions (carbonaut or something similar) being meh

[Falkentyne]

2 hours ago, mido said:

How well do the films work? I remember similar solutions (carbonaut or something similar) being meh

 

I haven't tested any of them yet.  Maybe I'll get around to it when i get a Raptor Lake CPU, since I'll have to switch to a 420mm AIO (liquid freezer) anyway, and I already have the Thermalright retention anti-bend CPU mounting mechanism anyway, using LM on IHS with LFII 360..

[Vasudev]

15 hours ago, mido said:

How well do the films work? I remember similar solutions (carbonaut or something similar) being meh

Depends on heatsink fit. On my alienware the GPU side had slight offset to the left while removing the heatsink. I had to redo it and tape on sides. On CPU side there were no slips. Then again I did cut the film/carbon sheet smaller as to fit it inside the CPU or GPU.

[Mr. Fox]

22 hours ago, mido said:

How well do the films work? I remember similar solutions (carbonaut or something similar) being meh

I tried using an Indium foil sheet, and even on my lapped IHS and perfectly flat water block the temps were higher than using ordinary thermal paste. I even ran an extended load test to "melt" the Indium foil and still wasn't good enough, and the foil either never melted or if it did it turned solid again after cooling and had no evidence of melting. It looked the same as when I inserted it and there was never any good result with it. Graphite sheets were even worse compared to the Indium.

[KING19]

Small update on Honeywell 7950SP paste:

 

It started to degrade somewhat because i did a run with cinebench and the CPU temps went as high as 98C at the max and while playing cyberpunk 2077 it hangs around the upper 70Cs to low 80Cs with spikes up to 91c. Other games stays around the 70c range with spikes to the upper 80Cs at times. I think because its cured now and plus its officially summer here on the east coast so its been hot and humid lately. Personally im not going to worry about it until i see 100C temps while playing GTA V or even American Truck Simulator like last time. Btw my GPU temps remains in the upper 60c's so no change there

[jaybee83]

17 hours ago, KING19 said:

Small update on Honeywell 7950SP paste:

 

It started to degrade somewhat because i did a run with cinebench and the CPU temps went as high as 98C at the max and while playing cyberpunk 2077 it hangs around the upper 70Cs to low 80Cs with spikes up to 91c. Other games stays around the 70c range with spikes to the upper 80Cs at times. I think because its cured now and plus its officially summer here on the east coast so its been hot and humid lately. Personally im not going to worry about it until i see 100C temps while playing GTA V or even American Truck Simulator like last time. Btw my GPU temps remains in the upper 60c's so no change there

important thing when checking thermal paste and related temps is the ambient temp. ull always need to calculate the delta between component temps and ambient temps. for every additional degree ambient ull get an additional degree heat on all your components, unfortunate fact.

[cylix]

On 7/3/2022 at 9:10 PM, jaybee83 said:

be sure that this is not due to poorly fitted heatsink vs actual thermal paste quality. liquid metal in the former case would do absolutely nothing and give u worse temps than any regular paste, actually... perfect fit is key in that regard 

 Yeah, ill need to see how straight is my heatsink and probably ill use k-5 pro instead of thermal pads for better pressure. Is electrical tape enough to isolate the board from LM spills?

[jaybee83]

20 minutes ago, cylix said:

 Yeah, ill need to see how straight is my heatsink and probably ill use k-5 pro instead of thermal pads for better pressure. Is electrical tape enough to isolate the board from LM spills?

if the fit is perfect and u can apply only barely enough to "wetten" the die and respective heatsink surface, then yes. if not ull want to look into foam dams surrounding the dies

[KING19]

On 7/5/2022 at 12:14 PM, jaybee83 said:

important thing when checking thermal paste and related temps is the ambient temp. ull always need to calculate the delta between component temps and ambient temps. for every additional degree ambient ull get an additional degree heat on all your components, unfortunate fact.

 

Yea you're right. Like i said its been hot and humid here lately because its summer, Damn east coast weather. We have an AC in our apartment but its off for most of the day because me and my GF are at work for most of the day. In my last gaming session the temps are still the same so i dont think its the paste itself anymore but we'll see after another week. It rained here for the last couple of days but the air is still humid. Our ambient temps is around 80f (26.6C) with the AC off but drops down to 74F (23.C) after running it for a couple of hours.

[ryan]

pretty cool thread...:)

 

just wondering if theres anything that performs close to liquid metal but is not conductive

[jaybee83]

5 hours ago, ryan said:

pretty cool thread...:)

 

just wondering if theres anything that performs close to liquid metal but is not conductive

direct die water cooling 😄 

[JeanLegi]

Who need LM or Paste when you can do something like this:

https://www.pugetsystems.com/aquarium_computer/buy.php

 

[jaybee83]

5 hours ago, JeanLegi said:

Who need LM or Paste when you can do something like this:

https://www.pugetsystems.com/aquarium_computer/buy.php

 

oh yeah, um....could u please pull out the cpu or gpu real quick? wanna check something.....yeah, sure, ill wait A WEEK TO LET THE COMPONENTS FULLY DRIP OFF! 😂

[i.bakar]

well since i was asking the exact same questions earlier ( i'm still are )

here is my experience with some of the above mentioned thermal pastes 

- alienware M18X R1

1- used mx4 on the 2960xm on the m18x r1 for 1 week,  temps was not manageable with 4.4ghz overclock and 1.4V ( temps was at 100c and shutdown after less then 3 minutes of rendering on vray 3ds max CPU render ( similar to what cinebench does )  

2- replaced the above with IC diamond on the CPU and left it there for a year.  temps  went to 91c for 10 minutes rendering on vray 3ds max CPU render ( similar to what cinebench does )  (a t the first week ) 

after 9 months of using it. the temps on the CPU increased by almost 3 degrees C 

3- replaced the ic diamond with  kryonaut on the CPU (currently using for the past 3 months ) temps  went to 95c for 10 minutes rendering on vray 3ds max CPU render ( similar to what cinebench does )

4- used mx4 on the 880M sli. for 1 week while rendering on vray 3ds max GPU render ( little less load then gaming )  temps was maxed and  stable at 78C , however it took it 4 minutes to reach this temp and stay there. 

5- used ic diamonds on the GPU and left it there for a year temps maxed and stabled at 73C   however it took it 10 minutes to reach this temp and stay there. 

6- replaced the ic diamond with  kryonaut on the GPU (currently using for the past 3 months ) temps didnt change at all from the IC diamond stabled at 73C  but it took abit less time to reach there 

 

- alienware 18 

1- used IC diamond on the CPU  4930mx over clocked to 4.3ghz with 1.27V and left it there for a year temps ware not manageable on  rendering on vray 3ds max CPU render ( similar to what cinebench does )   and shutdown due to the  thermals so went back to 4.2ghz and 1.18V and temps were never stable also but could get away with 15 minutes rendering without the laptop shuts down .

2- used liquid metal conductonaut on the same CPU for 4 months and at the begaining the temps was stable with the same 4.2 overcloecking at 1.18V at 85C in the same rendering mentioned above for almost as far as my most of my slw rendering ( 1 hour ) however after about 3 monthes of use i started to notice that the shutdown issues started to come up again due to the thermals and i had to repaste after 4 months with liquid metal again and same thing happed after anothe 4 months. 

3- used ic diamond on the GPU ( same 880M sli that i have now on the m18x r1) temps were stable at the exact same spot on the m18x r1  73C  ( using the same heatsinck on both laptops ) from the alienware 18

 

- alienware area 51-m

1- laptop came with paste from the factory on the CPU  9900k so the temps were stable at 91C with 4.6GHZ all cores at 1.2V at the same vray rendering as mentioned above. and stable at 96C with 4.7ghz all cores at 1.23V ( using only alienware command center )

2- replaced the above unknown paste with  kryonaut on the CPU so the temps were stable at 95C with 4.6GHZ all cores at 1.21V at the same vray rendering as mentioned above. and NOT stable at 4.7ghz all cores at 1.23V ( using only alienware command center ) so i had to go back to 4.6ghz 

3-  laptop came with paste from the factory on the GPU  RTX 2080 core clock overclocked 100mhz extra then stock and memory speed extra 200mhz temps were stable at 71C no throttling ( using the 180w maximum bios which is the new vbios update)  no power throttling too 

4- replaced the above unknown paste with  kryonaut on the GPU so the temps were stable at 75C with the same overclocking setting mentioned above , no power throttling but it thermal throttles !!

5- then with the same kryonaut   i used the 1.5 bios from @ssj92  and the older Vbios for the GPU which max at 200w and increased the thermal limit to 79C , the temps went to 79C then started to throttle thermally again which means that the bios and the Vbios helped to delay the throttling but the thermal paste would not decrease the temps by any means. 

that means that the dell stock thermal paste was better then the Kryonaut  from my experiance!

 

so the question now is , what thermal paste i should try next on the area-51M since so far the IC diamond was the best performer on the alienware m18x r1 and the alienware 18 ? 

 

[Eban]

21 hours ago, JeanLegi said:

Who need LM or Paste when you can do something like this:

https://www.pugetsystems.com/aquarium_computer/buy.php

 

Man that is beautiful!

A real pity about that patent garbage

 

[ssj92]

On 7/9/2022 at 3:02 AM, i.bakar said:

well since i was asking the exact same questions earlier ( i'm still are )

here is my experience with some of the above mentioned thermal pastes 

- alienware M18X R1

1- used mx4 on the 2960xm on the m18x r1 for 1 week,  temps was not manageable with 4.4ghz overclock and 1.4V ( temps was at 100c and shutdown after less then 3 minutes of rendering on vray 3ds max CPU render ( similar to what cinebench does )  

2- replaced the above with IC diamond on the CPU and left it there for a year.  temps  went to 91c for 10 minutes rendering on vray 3ds max CPU render ( similar to what cinebench does )  (a t the first week ) 

after 9 months of using it. the temps on the CPU increased by almost 3 degrees C 

3- replaced the ic diamond with  kryonaut on the CPU (currently using for the past 3 months ) temps  went to 95c for 10 minutes rendering on vray 3ds max CPU render ( similar to what cinebench does )

4- used mx4 on the 880M sli. for 1 week while rendering on vray 3ds max GPU render ( little less load then gaming )  temps was maxed and  stable at 78C , however it took it 4 minutes to reach this temp and stay there. 

5- used ic diamonds on the GPU and left it there for a year temps maxed and stabled at 73C   however it took it 10 minutes to reach this temp and stay there. 

6- replaced the ic diamond with  kryonaut on the GPU (currently using for the past 3 months ) temps didnt change at all from the IC diamond stabled at 73C  but it took abit less time to reach there 

 

- alienware 18 

1- used IC diamond on the CPU  4930mx over clocked to 4.3ghz with 1.27V and left it there for a year temps ware not manageable on  rendering on vray 3ds max CPU render ( similar to what cinebench does )   and shutdown due to the  thermals so went back to 4.2ghz and 1.18V and temps were never stable also but could get away with 15 minutes rendering without the laptop shuts down .

2- used liquid metal conductonaut on the same CPU for 4 months and at the begaining the temps was stable with the same 4.2 overcloecking at 1.18V at 85C in the same rendering mentioned above for almost as far as my most of my slw rendering ( 1 hour ) however after about 3 monthes of use i started to notice that the shutdown issues started to come up again due to the thermals and i had to repaste after 4 months with liquid metal again and same thing happed after anothe 4 months. 

3- used ic diamond on the GPU ( same 880M sli that i have now on the m18x r1) temps were stable at the exact same spot on the m18x r1  73C  ( using the same heatsinck on both laptops ) from the alienware 18

 

- alienware area 51-m

1- laptop came with paste from the factory on the CPU  9900k so the temps were stable at 91C with 4.6GHZ all cores at 1.2V at the same vray rendering as mentioned above. and stable at 96C with 4.7ghz all cores at 1.23V ( using only alienware command center )

2- replaced the above unknown paste with  kryonaut on the CPU so the temps were stable at 95C with 4.6GHZ all cores at 1.21V at the same vray rendering as mentioned above. and NOT stable at 4.7ghz all cores at 1.23V ( using only alienware command center ) so i had to go back to 4.6ghz 

3-  laptop came with paste from the factory on the GPU  RTX 2080 core clock overclocked 100mhz extra then stock and memory speed extra 200mhz temps were stable at 71C no throttling ( using the 180w maximum bios which is the new vbios update)  no power throttling too 

4- replaced the above unknown paste with  kryonaut on the GPU so the temps were stable at 75C with the same overclocking setting mentioned above , no power throttling but it thermal throttles !!

5- then with the same kryonaut   i used the 1.5 bios from @ssj92  and the older Vbios for the GPU which max at 200w and increased the thermal limit to 79C , the temps went to 79C then started to throttle thermally again which means that the bios and the Vbios helped to delay the throttling but the thermal paste would not decrease the temps by any means. 

that means that the dell stock thermal paste was better then the Kryonaut  from my experiance!

 

so the question now is , what thermal paste i should try next on the area-51M since so far the IC diamond was the best performer on the alienware m18x r1 and the alienware 18 ? 

 

Try Gelid GC-Extreme , @Mr. Fox recommended it to me awhile back. Surprisingly the all burning furnace GTX 880M is maxing out at 70C for me in M17x R3 using this. Also working with 2920XM in M17xR3 stock heatsink low 70s stock. 

 

Not sure how long it'll last but so far been almost a month. 

[i.bakar]

58 minutes ago, ssj92 said:

Try Gelid GC-Extreme , @Mr. Fox recommended it to me awhile back. Surprisingly the all burning furnace GTX 880M is maxing out at 70C for me in M17x R3 using this. Also working with 2920XM in M17xR3 stock heatsink low 70s stock. 

 

Not sure how long it'll last but so far been almost a month. 

i was actually between the GC extreme and the kingpin cooling KPX and i watched alot of comparison videos between them and the kryonaut, from what i saw so far the GS extreme and the KPX is almost head to head underload

 

 

so yesterday i pulled the trigger on the KPX. but  thanks for the recommendation. 

i started doing some cenabenck R23 tests and took a few screenshots for the score results along with the temperatures and the volts of the CPU and once i receive the KPX i will do the same tests and will share what i collected here for a more clearer comparison between the KPX and the kryonaut. 

 

[Mr. Fox]

The problem I had with Gelid and NTH1 is that both were inadequate viscosity and not durable due to pump-out. Kryosnaut was marginally better. All worked close to the same when first applied, but didn't last long enough. NTH1 was the least durable.

 

I have very good results with KPX on liquid cooled desktops, but not sure how durable it would be on a laptop.

 

Someone should try Thermalright TF7.  I used it for an experiment on one of my desktops to avoid wasting KPX and it worked quite well. Significantly better than I expected it to work. A tube of TF7 was included each of my 12th Gen Thermalright CPU frames. 

 

The TF7 was easy to apply, but I noticed when I removed it a couple of weeks later it had thickened a lot and was clay-like. It resisted wiping off the CPU and water block as easily as most thermal pastes do. It took at least twice as much effort to remove. It was not as thick as old IC Diamond, but impressively resistant to wiping off with an alcohol pad. My impression from that is that it might be very good on a laptop. Based on what I saw, I doubt it would be as susceptible to pump-out as most of the popular options.

[JeanLegi]

The KPX is currently not available same like IC Diamond in europe...

Buying from the US 60$ and shipment 30$... That's a bit to much money to spent 😄

[Mr. Fox]

3 hours ago, JeanLegi said:

The KPX is currently not available same like IC Diamond in europe...

Buying from the US 60$ and shipment 30$... That's a bit to much money to spent 😄

Yeah, Brother @JeanLegithat is pretty crazy. If it was proven to last as long as liquid metal (llasting years if applied in the proper conditions) and you were not frequently tinkering with things it might be worth it. For most that would be way too much money. 

 

As we all know, the problem with laptops is, what works good for desktop with parts that fit nice might be horrible for a laptop (because they are built so sloppy and parts don't fit nicely) and not last very long even if it works well at first application. And, what might do a nice job for a laptop might not be as good as something else in a desktop application. It can be very costly and time consuming playing the thermal paste lottery trying to find out what works best on a laptop. When you find it, even then you can only hope that it works as well on another laptop... it might not.

[1610ftw]

10 hours ago, Mr. Fox said:

The problem I had with Gelid and NTH1 is that both were inadequate viscosity and not durable due to pump-out. Kryosnaut was marginally better. All worked close to the same when first applied, but didn't last long enough. NTH1 was the least durable.

 

I have very good results with KPX on liquid cooled desktops, but not sure how durable it would be on a laptop.

 

Someone should try Thermalright TF7.  I used it for an experiment on one of my desktops to avoid wasting KPX and it worked quite well. Significantly better than I expected it to work. A tube of TF7 was included each of my 12th Gen Thermalright CPU frames. 

 

The TF7 was easy to apply, but I noticed when I removed it a couple of weeks later it had thickened a lot and was clay-like. It resisted wiping off the CPU and water block as easily as most thermal pastes do. It took at least twice as much effort to remove. It was not as thick as old IC Diamond, but impressively resistant to wiping off with an alcohol pad. My impression from that is that it might be very good on a laptop. Based on what I saw, I doubt it would be as susceptible to pump-out as most of the popular options.

 

Got a tube of TF8 here and may try that soon. According to Thermalright there only is a gradual increase in effectiveness from TF7 to TFX that goes as follows:

 

TF7: 12.8W/mK

TF8: 13.8W/mK

TFX: 14.3W/mK

 

The biggest issue is of course that in laptops nothing is ever easy and straight so that lots of time has to be spent on perfecting what is inherently very far from perfect.

 

[i.bakar]

14 hours ago, Mr. Fox said:

The problem I had with Gelid and NTH1 is that both were inadequate viscosity and not durable due to pump-out. Kryosnaut was marginally better. All worked close to the same when first applied, but didn't last long enough. NTH1 was the least durable.

 

I have very good results with KPX on liquid cooled desktops, but not sure how durable it would be on a laptop.

 

Someone should try Thermalright TF7.  I used it for an experiment on one of my desktops to avoid wasting KPX and it worked quite well. Significantly better than I expected it to work. A tube of TF7 was included each of my 12th Gen Thermalright CPU frames. 

 

The TF7 was easy to apply, but I noticed when I removed it a couple of weeks later it had thickened a lot and was clay-like. It resisted wiping off the CPU and water block as easily as most thermal pastes do. It took at least twice as much effort to remove. It was not as thick as old IC Diamond, but impressively resistant to wiping off with an alcohol pad. My impression from that is that it might be very good on a laptop. Based on what I saw, I doubt it would be as susceptible to pump-out as most of the popular options.

NTH1 is the least durable from what i hear too . however i already ordered KPX so i will be testing it once i receive it within 20 days hopefully. lets see how it will perform on the 9900k against the Kryosnaut .

 

[i.bakar]

4 hours ago, Mr. Fox said:

Yeah, Brother @JeanLegithat is pretty crazy. If it was proven to last as long as liquid metal (llasting years if applied in the proper conditions) and you were not frequently tinkering with things it might be worth it. For most that would be way too much money. 

 

As we all know, the problem with laptops is, what works good for desktop with parts that fit nice might be horrible for a laptop (because they are built so sloppy and parts don't fit nicely) and not last very long even if it works well at first application. And, what might do a nice job for a laptop might not be as good as something else in a desktop application. It can be very costly and time consuming playing the thermal paste lottery trying to find out what works best on a laptop. When you find it, even then you can only hope that it works as well on another laptop... it might not.

i dont know if i should try liquid metal on the 9900k on the area51m i still have some leftover . the issue is that i dont wanna deled it so is it possible to do liquid metal between the die and the heatsink like what we do on the laptop processors ? what so u think ?

[Mr. Fox]

On 7/11/2022 at 10:59 AM, i.bakar said:

i dont know if i should try liquid metal on the 9900k on the area51m i still have some leftover . the issue is that i dont wanna deled it so is it possible to do liquid metal between the die and the heatsink like what we do on the laptop processors ? what so u think ?

Yes, absolutely. I do this all of the time, even if the CPU is not delidded. If the contact between the IHS and heat sink is good the liquid metal will produce much better results.  If you travel with the laptop or move it around a lot, carry it in a backpack, etc. then you should consider building a barrier/dam around the perimeter of the CPU with some thin foam and use Kapton tape to bridge the space between the ILM (CPU retainer) and IHS. If you do not use an excessive amount of liquid metal and do not ragdoll the laptop it is not super dangerous, but the dam is cheap insurance to avoid a tragic mishap.

 

The foam needs to be very thin and easily compressed to avoid causing contact interference. I used a few drops of SuperGlue on the Kapton tape to hold the thin outer foam dam in place while assembling things. The inner foam dam under the ILM is held in place by both the ILM and the Kapton tape.

 

These photos are from the X170 cooling enhancements I did for zTecpc. I know that @electrosoftand @Clamibot are owners of that machine, but you can probably adopted some of this for your Alienware.

[i.bakar]

49 minutes ago, Mr. Fox said:

Yes, absolutely. I do this all of the time, even if the CPU is not delidded. If the contact between the IHS and heat sink is good the liquid metal will produce much better results.  If you travel with the laptop or move it around a lot, carry it in a backpack, etc. then you should consider building a barrier/dam around the perimeter of the CPU with some thing foam and use Kapton tape to bridge the space between the ILM (CPU retainer) and IHS. If you do not use an excessive amount of liquid metal and do not ragdoll the laptop it is not super dangerous, but the dam is cheap insurance to avoid a tragic mishap.

 

The foam needs to be very thin and easily compressed to avoid causing contact interference. I used a few drops of SuperGlue on the Kapton tape to hold the thin outer foam dam in place while assembling things. The inner foam dam under the ILM is held in place by both the ILM and the Kapton tape.

 

These photos are from the X170 cooling enhancements I did for zTecpc. I know that @electrosoftand @Clamibot are owners of that machine, but you can probably adopted some of this for your Alienware.

big thanks for the idea and the pictures , that will make my job much easier