Ok guys thanks for the replies. Ive ordered some liquid metal, it should turn up wed/thurs. So i'll report back my findings and see what happens!
8700k delid made no difference at all?
- Thread starter C6ckneyGeezer
- Start date
Well I have stripped it all down again as my Thermal Grizzly Conductonaut turned up! As you can see from the pictures I cleaned both the CPU die, IHS and NH-D14 cooler, firstly with tissue and cotten buds etc to the get the worst of the paste off, then with the supplied alcohol wipes.
I taped up both the CPU die and IHS with tape, then proceeded to spread the Conductonaut. I chose to apply a very very small pin head, but then needed a little more. However to add the little bit extra, I squirted some onto the packet and dabbed the supplied bud onto it, so I didn't add too much. It comes out really fast!
I then removed all the tape and applied some nail varnish to the small terminals near the die. This insured no Conductonaut was going to short the terminals. It was then left alone for about 20 mins for the nail varnish to fully harden before I installed the CPU into the socket and then secured the IHS to it. I didn't choose to reseal as I wanted to test my results. And tbh, I think i'll leave it how it is.
I then applied some Thermal Grizzly Kryonaut between the IHS and my NH-D14 cooler, using a "small sized pea" method.
The results!
After over 30 mins of NON AVX Prime (Version 26.6) these are my max temps! This is with an ambient of 22*c, reads 24*c inside my case. Currently at 4.6Ghz @ 1.190v with llc set to 6, which gives me 1.200v under load.
Package - 65*c
Core0 - 63*c
Core1 - 61*c
Core2 - 65*c
Core 3 - 62*c
Core 4 - 61*c
Core 5 - 61*c
Well...wow. That is ineeded more than 20*c off my temps! My temps before the Conductonaut were mid to high 80's follows with an ambient of 20*c.
Well happy! Thanks everybody! Now time to push that OC I guess
Edit: Oh I forgot the results for AVX Prime (Version 29.4 latest as of this post)
Package - 73*c / +8*c compared NON AVX Prime
Core0 - 71*c
Core1 - 69*c
Core2 - 73*c
Core 3 - 70*c
Core 4 - 69*c
Core 5 - 69*c
I taped up both the CPU die and IHS with tape, then proceeded to spread the Conductonaut. I chose to apply a very very small pin head, but then needed a little more. However to add the little bit extra, I squirted some onto the packet and dabbed the supplied bud onto it, so I didn't add too much. It comes out really fast!
I then removed all the tape and applied some nail varnish to the small terminals near the die. This insured no Conductonaut was going to short the terminals. It was then left alone for about 20 mins for the nail varnish to fully harden before I installed the CPU into the socket and then secured the IHS to it. I didn't choose to reseal as I wanted to test my results. And tbh, I think i'll leave it how it is.
I then applied some Thermal Grizzly Kryonaut between the IHS and my NH-D14 cooler, using a "small sized pea" method.
The results!
After over 30 mins of NON AVX Prime (Version 26.6) these are my max temps! This is with an ambient of 22*c, reads 24*c inside my case. Currently at 4.6Ghz @ 1.190v with llc set to 6, which gives me 1.200v under load.
Package - 65*c
Core0 - 63*c
Core1 - 61*c
Core2 - 65*c
Core 3 - 62*c
Core 4 - 61*c
Core 5 - 61*c
Well...wow. That is ineeded more than 20*c off my temps! My temps before the Conductonaut were mid to high 80's follows with an ambient of 20*c.
Well happy! Thanks everybody! Now time to push that OC I guess
Edit: Oh I forgot the results for AVX Prime (Version 29.4 latest as of this post)
Package - 73*c / +8*c compared NON AVX Prime
Core0 - 71*c
Core1 - 69*c
Core2 - 73*c
Core 3 - 70*c
Core 4 - 69*c
Core 5 - 69*c
Last edited:
I had re done it all about 4 times before hand with paste when delidding. Checked everything over and over. First go with Conductonaut and that's the results!
Currently at 5.0Ghz @ 1.330v which gives me 1.344v under load using LLC Level 6. Seen a max temp of 77*c with NON AVX Prime. Also been running various benches and stress testing all day. Seems stable so far...
I'm glad you now have good heat transfer and great temps, but the reality is metal TIM is only 2-3c better than good name-brand normal TIM.
Here is link to testing of 85 TIMs including one you are using. There are 14 TIMs within 2c and 39 within 3c. Even using denture cream their temps were only 8c warmer than Conductonaut.
https://www.tomshardware.co.uk/thermal-paste-comparison,review-33969-8.html
Here is link to testing of 85 TIMs including one you are using. There are 14 TIMs within 2c and 39 within 3c. Even using denture cream their temps were only 8c warmer than Conductonaut.
https://www.tomshardware.co.uk/thermal-paste-comparison,review-33969-8.html
orbitalwalsh explains it correctly plus there is one more thing that make the conductonaut becomes super effective compared to to regular good tim in such case, that is because the gap between the IHS and the CPU core is incredibly close after you clean up all the black glue and with the mounting pressure "big" particles of regular tim will be push out or do not fill the micro gaps very well, on the other hand, conductonaut is made from mixing indium, gallium, tin, etc... which have significant smaller sized particles thus filling the micro gaps between the ihs and the actual core much more effective and due to this the heat transfer rate is superior compared to normal tim.I'm glad you now have good heat transfer and great temps, but the reality is metal TIM is only 2-3c better than good name-brand normal TIM.
Here is link to testing of 85 TIMs including one you are using. There are 14 TIMs within 2c and 39 within 3c. Even using denture cream their temps were only 8c warmer than Conductonaut.
https://www.tomshardware.co.uk/thermal-paste-comparison,review-33969-8.html
I have tried it myself several times and found similar results as the original poster.
The test you link is legit in relative to normal operation however it does not show to true potential of liquid metal based tims, these special tims works best only when the gap of the two surfaces becomes very very very close and the surfaces must be mirror-like finished (like the surface of the CPU cores when you clean up all the residuals). Then and only then it distances itself from regular tims.
Last edited:
orbitalwalsh explains it correctly plus there is one more thing that make the conductonaut becomes super effective compared to to regular good tim in such case, that is because the gap between the IHS and the CPU core is incredibly close after you clean up all the black glue and with the mounting pressure "big" particles of regular tim will be push out or do not fill the micro gaps very well, on the other hand, conductonaut is made from mixing indium, gallium, tin, etc... which have significant smaller sized particles thus filling the micro gaps between the ihs and the actual core much more effective and due to this the heat transfer rate is superior compared to normal tim.
I have tried it myself several times and found similar results as the original poster.
The test you link is legit in relative to normal operation however it does not show to true potential of liquid metal based tims, these special tims works best only when the gap of the two surfaces becomes very very very close and the surfaces must be mirror-like finished (like the surface of the CPU cores when you clean up all the residuals). Then and only then it distances itself from regular tims.
We do not want a layer of TIM between CPU and IHS, not even a thin layer. Sure, gallium is what gives liquid metal TIMs higher 73W/m k or similar rating. And sure, normal TIM is 6-12 W/m k and will raise CPU temps by about 2-5c .
But all TIM has much lower W/m k than a metal .. aluminum is 200 W/m k and copper is 400 W/m k. As a comparison air is only 0.024 W/m k. Regardless of how good the TIM is, the medium it is mixed into greatly lowers it's W/m k .. and a layer of even the best TIM will lower the W/m k significantly.
You've hit on nail of the proverbial head. We need to have the best metal to metal contact we can get with TIM only filling the microscopic voids in crystallize structure of metal. We do not want even a thin layer between the surfaces. The more metal to metal contact we have with TIM in void the better the heat transfer will be / the higher the W/m k. Problem is many IHS even after delidding can be give us a concave IHS surface meaning we end up with a layer of TIM between surface .. which is what I suspect C6ckneyGee3zer had happening with his other mountings.
Here's link to TIM application, W/m k ratings, how to get the best mount, etc.
http://www.overclock.net/forum/246-...ing-airflow-cooler-fan-data.html#post22335323
http://www.overclock.net/forum/attachment.php?attachmentid=182537&d=1525962932
Last edited:
Soldato
You're very right, but I have to point out that silicon isn't a metal
Neither is air
Granted the CPU is a silicon chip, but I was talking about transferring heat from that chip to IHS and on to cooler base/waterblock. TIM used and how it was applied / used is only things that changed in C6ckneyGee3zer's build with exaggerated claims about how great Conductonaut is when the maximum cooling increase between difference in TIMs used is 2-3c.
Granted the CPU is a silicon chip, but I was talking about transferring heat from that chip to IHS and on to cooler base/waterblock. TIM used and how it was applied / used is only things that changed in C6ckneyGee3zer's build with exaggerated claims about how great Conductonaut is when the maximum cooling increase between difference in TIMs used is 2-3c.
Im gona put it down to indeed the Conductonaut has knocked 20*c off my temps! I re seated, reapplied everything many many times before hand with consistant poor results. As soon as I used the Conductonaut instead of Kryonaut, things were exactly as I have been told to expect from advice and comments etc.
However. Like I said before, I did delid my 3770k using Kryonaut between the die and IHS and Kryonaut between the IHS and CPU cooler also with good results over stock of -20*c. Maybe from the 3770k era the stock thermal paste was indeed very poor? But with the 8700k not so poor? With Kryonaut not making any difference on the 8700k delid, maybe the stock paste is on par with Kryonaut? If so, the only benefit using Kryonaut over stock paste on a 8700k is loosing the IHS adhesive and creating a thinner bridge. Which made no reall difference.
By loosing the adhesive on the IHS and making it closer to the die, using a very very thin and more effective layer of heat transfer, Conductonaut....seems to of for sure done the trick.
Edit: Little update!
I seem to have settled with 4.8Ghz! 5Ghz gets a bit toasty!
4.8Ghz @ 1.265 = 1.280 under full load. LLC Level 6. Getting a max temp of 76*c with NON AVX Prime, 84*c with AVX Prime (Latest Build).
5.0Ghz @ 1.330 = 1.344 under full load. LLC Level 6. This see's me pushing into the 90's with AVX Prime. I'd rather run stable with AVX Prime, I know it's rock solid then. Could always run an AVX offset I guess as things were fine with NON AVX Prime.
I can't see there being any real world difference between 4.8 and 5.0 tbf...
It seems I have a decent clocker?
However. Like I said before, I did delid my 3770k using Kryonaut between the die and IHS and Kryonaut between the IHS and CPU cooler also with good results over stock of -20*c. Maybe from the 3770k era the stock thermal paste was indeed very poor? But with the 8700k not so poor? With Kryonaut not making any difference on the 8700k delid, maybe the stock paste is on par with Kryonaut? If so, the only benefit using Kryonaut over stock paste on a 8700k is loosing the IHS adhesive and creating a thinner bridge. Which made no reall difference.
By loosing the adhesive on the IHS and making it closer to the die, using a very very thin and more effective layer of heat transfer, Conductonaut....seems to of for sure done the trick.
Edit: Little update!
I seem to have settled with 4.8Ghz! 5Ghz gets a bit toasty!
4.8Ghz @ 1.265 = 1.280 under full load. LLC Level 6. Getting a max temp of 76*c with NON AVX Prime, 84*c with AVX Prime (Latest Build).
5.0Ghz @ 1.330 = 1.344 under full load. LLC Level 6. This see's me pushing into the 90's with AVX Prime. I'd rather run stable with AVX Prime, I know it's rock solid then. Could always run an AVX offset I guess as things were fine with NON AVX Prime.
I can't see there being any real world difference between 4.8 and 5.0 tbf...
It seems I have a decent clocker?
Last edited:
Yeah, elephants and pigs fly too.
I don't care how many times you re-did TIM on chip and iHS, you was not done correctly until this last time. It it had been done correctly your previous temps would have been about 2-3c warmer than they are not with Conductonaut .. 4-5c improvement is be a real stretch even with Conductonaut on both CPU and IHS.
My guess is you finally got all the caulk/glue off of IHS and CPU PCB so edges of IHS were flat with IHS over CPU chip, so instead of a layer of TIM between CPU chip and IHS compounded by another layer of TIM between IHS and cooler base you finally got direct silcon to metal with TIM filling void on chip to IHS and metal to metal with TIM filling voids between IHS and cooler.
I don't care how many times you re-did TIM on chip and iHS, you was not done correctly until this last time. It it had been done correctly your previous temps would have been about 2-3c warmer than they are not with Conductonaut .. 4-5c improvement is be a real stretch even with Conductonaut on both CPU and IHS.
My guess is you finally got all the caulk/glue off of IHS and CPU PCB so edges of IHS were flat with IHS over CPU chip, so instead of a layer of TIM between CPU chip and IHS compounded by another layer of TIM between IHS and cooler base you finally got direct silcon to metal with TIM filling void on chip to IHS and metal to metal with TIM filling voids between IHS and cooler.
doyll, have you ever experimenting with conductonaut/liquid metal versus regular kryonaut tim between IHS and CPU core? I have and weirdly my results are very similar with C6ckneyGee3zer's.
I have tried a considered number of times not only on my 8700Ks but at least a dozen 6700K/7700K in the past and found similar results.
I was trying to build a pre-overclocked for my long distance relationship gf, my intention was trying to provide a decently high clocked cpu (4.6Ghz quad core) with minimal maintenance since we could only meet once or twice a year hence my idea was deliding the cpu, then use regular tim like Kryonaut to replace the Intel stock tim since it is less risky compared to liquid metal. However, when I actually tried it I found it was still very hot (above 80) during load which made overclocking seems impossible, no matter how many times I tried to re-seat and re-paste it, the temp did not seem to get anywhere close to liquid metal. This somewhat verifies or at least gives one more similar result with what C6ckneyGee3zer has found with his CPU.
I have tried a considered number of times not only on my 8700Ks but at least a dozen 6700K/7700K in the past and found similar results.
I was trying to build a pre-overclocked for my long distance relationship gf, my intention was trying to provide a decently high clocked cpu (4.6Ghz quad core) with minimal maintenance since we could only meet once or twice a year hence my idea was deliding the cpu, then use regular tim like Kryonaut to replace the Intel stock tim since it is less risky compared to liquid metal. However, when I actually tried it I found it was still very hot (above 80) during load which made overclocking seems impossible, no matter how many times I tried to re-seat and re-paste it, the temp did not seem to get anywhere close to liquid metal. This somewhat verifies or at least gives one more similar result with what C6ckneyGee3zer has found with his CPU.
Yeah, elephants and pigs fly too.
I don't care how many times you re-did TIM on chip and iHS, you was not done correctly until this last time. It it had been done correctly your previous temps would have been about 2-3c warmer than they are not with Conductonaut .. 4-5c improvement is be a real stretch even with Conductonaut on both CPU and IHS.
My guess is you finally got all the caulk/glue off of IHS and CPU PCB so edges of IHS were flat with IHS over CPU chip, so instead of a layer of TIM between CPU chip and IHS compounded by another layer of TIM between IHS and cooler base you finally got direct silcon to metal with TIM filling void on chip to IHS and metal to metal with TIM filling voids between IHS and cooler.
I can assure you all the glue etc was off cleaned off properly. Everything was seated correctly. After re trying everything again 4-5 times with the same results...then I try Conductonaut and it works as it should..must be conductonaut.
doyll, have you ever experimenting with conductonaut/liquid metal versus regular kryonaut tim between IHS and CPU core? I have and weirdly my results are very similar with C6ckneyGee3zer's.
I have tried a considered number of times not only on my 8700Ks but at least a dozen 6700K/7700K in the past and found similar results.
I was trying to build a pre-overclocked for my long distance relationship gf, my intention was trying to provide a decently high clocked cpu (4.6Ghz quad core) with minimal maintenance since we could only meet once or twice a year hence my idea was deliding the cpu, then use regular tim like Kryonaut to replace the Intel stock tim since it is less risky compared to liquid metal. However, when I actually tried it I found it was still very hot (above 80) during load which made overclocking seems impossible, no matter how many times I tried to re-seat and re-paste it, the temp did not seem to get anywhere close to liquid metal. This somewhat verifies or at least gives one more similar result with what C6ckneyGee3zer has found with his CPU.
It's strange though as my delid results with my old 3770k were great, I lost 20*c when using Kryonaut. Maybe Intel having made their paste better since then? As there is not much difference between stock and aftermarket poaste on a delid it seems on modern CPU's?
Last edited:
I can understand 10c or better for pre-delidding to delidded. What I cannot understand is how simply changing pasts on an already delidded system because TIM does not make that great a difference. Delidding's biggest improvement is not better TIM but removing the space between CPU chip and IHS so there is direct contact between them instead of a layer of TIM. 10c is the kinds of difference seen between something like toothpaste and liquid metal TIM with direct contact between surfaces, or TIM between not contacting surfaces, not good TIM and liquid mental with direct contact between surfaces.
Maybe there was just too much paste...who knows...
What paste did you use?
Thanks for the initial help etc
Jury’s out on this one I suppose, only experience I have with it is a poorly applied layer of LM that was causing high temp spikes and higher average temps by about 10 degrees. Not quite the same thing but goes to show a less than 100% perfect interface between IHS and die can cause some dramatic results.
I might experiment with it next time I delid a PS3 but that might not give the same kinds of results as the die is much larger.
I might experiment with it next time I delid a PS3 but that might not give the same kinds of results as the die is much larger.
the layer of tim needs to be ultra thin like tinfoil, if you see any pooling its too thick..however if your ihs is warped then its not going to make any difference and maybe one of the copper ihs may be the solution,i was lucky I guess I checked my his and it was very flat.
I didn't use any sealant and let the ihs just float on the die and its locked down with the MB clamp.
I didn't use any sealant and let the ihs just float on the die and its locked down with the MB clamp.