3950x watercooling temperatures

[SparkleDJackson]

Ive been away from home the past few days for work but i will be going back home in The next few days so im probably going to check the block mounting and paste distribution and see if maybe i didnt use enough paste which is possible due to my inexperience with smaller blocks i may have under compensated i. Comparison to my 2011 blocks ive had for the past 8 years or however long it was the x79 came out. Ill update woth any findings i may have. Also. Ek plating is still terrible. My 2080ti block is flaking a little i drained out the loop and then did multiple flushes and was finding very fine silver powder clumps in the waste collector. And it is visually degraded. This isnt the case for the byksi block i had given the same pre fill cleaning and same fluids and same draining procedure and the bykski block that was in it before i had to do the rebuild is 100% flawless.

 

[SparkleDJackson]

So it seems like i used just a hair too little paste. Im going to re apply fresh paste and hope for the best. Had a giggle last night as my o.h.m. Showed my max package temp was 260c. the iimmedate temps listed below are over a 10 second period because it fluctuates like mad. its does seem to idle a few degrees lower though than before after reseating with more paste

| +- CPU Package immediate 42-50 max 53.3
| +- CPU CCD Max immediate 38-48 max 49.3
| +- CPU CCD Average immediate 36-42 max 45.3
| +- CPU CCD #1 immediate 37-42 max 49.3
| +- CPU CCD #2 immediate 35-39 max 46.8

i use open hardware monitor to show my stats if there is any better programs that also show cpu and gpu temps on a desktop widget that is more acccurate than this app please let me know

 

[Avet]

So it seems like i used just a hair too little paste. Im going to re apply fresh paste and hope for the best. Had a giggle last night as my o.h.m. Showed my max package temp was 260c. the iimmedate temps listed below are over a 10 second period because it fluctuates like mad. its does seem to idle a few degrees lower though than before after reseating with more paste

| +- CPU Package immediate 42-50 max 53.3
| +- CPU CCD Max immediate 38-48 max 49.3
| +- CPU CCD Average immediate 36-42 max 45.3
| +- CPU CCD #1 immediate 37-42 max 49.3
| +- CPU CCD #2 immediate 35-39 max 46.8

i use open hardware monitor to show my stats if there is any better programs that also show cpu and gpu temps on a desktop widget that is more acccurate than this app please let me know

The focused heat on AMD CPUs isn't at the center of the die like it is with Intel. You need to spread it well buy hand all over the IHS.

 

[SparkleDJackson]

Yea i manually spread the paste this time and after 20 runs of timespy stress test it spikes to 60 for a second or two at the beginning of each phase but then settles around 55 during the tests. So i think its helped for idle temps a bit but and the load temps are a little tighter so that little patch made a good difference lol. I know what to do when my 3900x arrives later today for the second system.

 

[Avet]

When you see guys like Der8auer spread their thermal paste and tell you that you should cover the whole IHS and not put just a droplet in the middle why would anyone argue differently?

 

[AlienALX]

When you see guys like Der8auer spread their thermal paste and tell you that you should cover the whole IHS and not put just a droplet in the middle why would anyone argue differently?

Because he is not the lord of science and physics. He spreads and uses his TIM in the way that he does because he is into extreme cooling.

As it goes Kryonaut is not the best TIM out there. It's just the only one being talked about and hyped up. Personally I thought it was lacklustre and completely over priced for regular use. The Kryo part kind of gives it away, it is for sub ambient cooling. Noctua is better for regular use. You also get more than one application out of Noctua too.

Science dictates that the TIM is used to fill the pits and voids between the IHS and cooler or block, and thus should be the thinnest it can be for optimal transfer.

There are plenty of TIMs costing about 1/3 of his stuff and work every bit as good or better for standard cooling.

 

[Warchild]

Because he is not the lord of science and physics. He spreads and uses his TIM in the way that he does because he is into extreme cooling.

As it goes Kryonaut is not the best TIM out there. It's just the only one being talked about and hyped up. Personally I thought it was lacklustre and completely over priced for regular use. The Kryo part kind of gives it away, it is for sub ambient cooling. Noctua is better for regular use. You also get more than one application out of Noctua too.

Science dictates that the TIM is used to fill the pits and voids between the IHS and cooler or block, and thus should be the thinnest it can be for optimal transfer.

There are plenty of TIMs costing about 1/3 of his stuff and work every bit as good or better for standard cooling.

This is why LM was so superior at one point. Its layer being thinner than any paste. But TG naming of Kryonaut is a little misleading I guess. They didnt design it for sub zero. Its just it works well at that temp. Although their own site, "suggests it was designed for crygenics". Kind of weird their page would suggest why it was designed. Seems TG are unsure themselves

I believe its actually their best all rounder. It was designed for extremely demanding applications (Their words).

I used and found it too thick for my liking, although it is one of the best I have used. Regardless though.. I always end up returning to LM. And this time I went for TG conductonaut over cool laboratory ultra. Was suprised to see 5C difference comparing both liquid metals. Not sure why we would see such a difference.

„Kryo" - the greek word for "cold" - also found in the german word for "cryoengineering", suggests that this thermal grease was created especially for cryogenic applications - for the true "Kryonauts" amongst the "extreme overclockers".

It's like they dont know for certain themselves

 

[Avet]

We can talk about science all day. More surface contact = more heat transfer. The pea method leaves ~20-30% of the IHS without contact with the cooler. This wasn't an issue before because Intel used toothpaste on the die. No matter how big the loop you use heat was trapped in the CPU. Also, the cooler often isn't good enough to show you the difference. If you would use 10980XE or 10900K with a 420mm or 480mm rad you would see a difference between spread and pea method.

Physics tells us that spreading thermal paste is better than the pea method. Every part of the IHS that isn't in contact with cooler doesn't transfer heat. If bottlenecks are somewhere else it nullifies the method used.

When you look at Threadripper good full cover Aircoolers will outperform the best Asetek based AIOs just because of the bigger contact surface.

Also on Ryzen CPUs heat source isn't at the center. It is shifted to the side over the chiplets. It defies the common sense to put a pea-sized amount of thermal paste in the middle.

Thermal Grizzly is the best thermal paste. It is only beaten by liquid metal tims. Here is the review with 85 products tested:

https://www.tomshardware.com/reviews/thermal-paste-comparison,5108-8.html

Edit: And also the thermal paste can contribute to no difference. If you would use liquid metal and do a circle in the middle vs full IHS with good enough cooler you would definitely see a difference.

 

[SparkleDJackson]

Ive never done the manual spread before because i usually used the x method for my x79/x99 builds so i just very slightly under used the paste. Its all gucci now. I just use mx4 now but i used to use ic diamond. But it was like 3 times the price and was a right whore to clean. I dont look at all the derbauers and 8 pcks because im not going to go to the extent of fiddling with micro volts or delid. I did notice the ihs appears to have a very slight dish. Maybe its for expansion.

 

[AlienALX]

I've not had the same results. That said paying four times as much for something that apparently and clearly makes such a tiny difference is nuts.

It's decent for direct to die on GPUs as it's not runny but I just don't rate it on CPUs.

Oh and yes, it's too thick also which is why you have to use so much. It's like silly putty ffs.

 

[Warchild]

I've not had the same results. That said paying four times as much for something that apparently and clearly makes such a tiny difference is nuts.

It's decent for direct to die on GPUs as it's not runny but I just don't rate it on CPUs.

Oh and yes, it's too thick also which is why you have to use so much. It's like silly putty ffs.

Thats not much different to what we are paying for GPU and CPUs for 10% or less upgrades. That is for both AMD and Intel, yet we still do it because... benchmarks

 

[Bartacus]

I'm with Avet, spreading FTW. I've always found spreading to be the most consistent way to apply paste, but YMMV. It's why I still prefer MX-4 too, since that stuff spreads very nicely.

 

[looz]

Only reason not to use liquid metal is preserving warranty IMO.

 

[AlienALX]

I'm with Avet, spreading FTW. I've always found spreading to be the most consistent way to apply paste, but YMMV. It's why I still prefer MX-4 too, since that stuff spreads very nicely.

Some pastes have to be spread. They are too thick to spread properly when you screw down the cooler. Case in point, Arctic Ceramique which I used for many years.

It depends on the paste and the viscosity.

Thats not much different to what we are paying for GPU and CPUs for 10% or less upgrades. That is for both AMD and Intel, yet we still do it because... benchmarks

Thing is if you are changing anything at all ever it just costs more and more money for one application. I wouldn't hesitate to use it on a GPU again, it's very good for that. However, as an example when I did an upgrade over Christmas I had to pull the CPU out four times and continually switch it back with the old one to troubleshoot.

I decided against it on my 3950x because I knew I was going to pull the cooler when my der8auer bracket finally arrived. Which it has, so I need to pull the cooler off and sort that soon. But that would have been another £8 down the drain.

 

[NeverBackDown]

Well it's worth it for some people. If you run a business that small tiny gain can still surmount to some form of revenue increase or allow for more complex services that are able to finish in the same amount of time as before. Which increases quality or accuracy. It scales with economies of scale and becomes more worthwhile. Just depends where that individual ends up on that scale.

For people like us though no it's not worth it. Unless you want it then of course that's your money I won't judge and if you're happy then who cares