Question Not able to cool i9-9900K with a 360mm CUSTOM LOOP!!!

May 12, 2019
32
1
30
Hi, I own an i9-9900K and I bought a 360mm radiator with a custom loop ONLY FOR THE CPU (EK Fluid gaming A360G) to cool it.
But I have really high temps at stock clocks: When runing Aida 64for 10 minutes, it throttles in a 5-10%range. With tems really high. CPU at: 83º, CPU core 1: 90ª, CPU c2: 75ª, core 3: 98ª cpu cre 4:
100ª e.t.c

I think these tems are really high for a custom 360mm loop. What do you think? What could be causing the issue???
 

rubix_1011

Contributing Writer
Moderator
What are your ambient room temps?

Are your idle temps (CPU 0%) between 20-32C or so?

After benching, do your temps drop dramatically back down to idle temps, or do they remain high?

Can you confirm the pump is moving coolant and setup correctly? Is pump PWM being used? Is the pump speed set low or at full speed?

What thermal compound is being used? Provided by EK in the kit, or other? Is the CPU block correctly and fully seated?
 
May 12, 2019
32
1
30
Hi,
Thanks for your answer. My idle tems are of 23º (CPU general) and between 30-32º in core 1,2 e.t.c.
Temps (After benchmark) drop dramatically back down to idle temps .

Yes, the pump is mobing coolant. Punp PWM is being used and the pump (when tems are above 60º) is at full speed.
The thermal compound is the one provided by EK and I think that the CPU block is correctly and fully seated.

Thanks.
 
Last edited:

oczdude8

Distinguished
Temps are high, I would expect around 50-70c depending on radiator fan and pump speeds

Did you ever run this CPU with a different cooler? What temps did you get?

Also when you start the benchmarks, does the CPU temp jump very quickly to high temps?

I can think of two things so far that it could be:
  1. Air bubble in the loop/CPU block
  2. Poor thermal paste application
 
May 4, 2019
59
6
45
What's the total package power under load?

Direction/orientation of the water block and loop - make sure the center is the input. If vertical, ensure that the output from the block is at the top. Check the loop for air by turning it different directions to see if any air comes out into the tubing from the block or radiator. Are the radiator fans the proper typ for the airflow direction.
 

rubix_1011

Contributing Writer
Moderator
Sounds like lack of flow is not the issue, how are you confirming pump speed? I'm not a fan of using PWM pump features...why not just set it at 100% or your desired flow and leave it? Is there obvious coolant circulation in your reservoir? It should be very noticeable in a loop like that.

There could be air in the radiator, hard to say, but if the loop is brand new, depending on where the radiator is located and oriented, this might or might not be an issue to check on.

If you shut down the computer and pull the CPU block, is the thermal paste spread evenly with a good contact patch?

Did you setup the EK CPU block with the correct jet direction and flow plate? (Does the CPU block in your kit have this option?)

What are your radiator fans set to at 100% load?

What case? How is it installed? Can you post a photo?
 
May 12, 2019
32
1
30
@mark1776
The total package power under load (when running Aida64 stressing cpu, FPU and cache) is ob 188W max.

"Direction/orientation of the water block and loop" IMAGE OF MY LOOP: https://drive.google.com/file/d/1cNs76kNV0j9YuIJhHNNSVgu-tbtrE_KX/view

I made sure the center is the input.
" If vertical, ensure that the output from the block is at the top." Can you please explain it in a different way, I could not underestand it?

"Check the loop for air by turning it different directions to see if any air comes out into the tubing from the block or radiator"

Are the radiator fans the proper typ for the airflow direction. " IMAGE OF MY LOOP: https://drive.google.com/file/d/1cNs76kNV0j9YuIJhHNNSVgu-tbtrE_KX/view
 
May 12, 2019
32
1
30
"Sounds like lack of flow is not the issue, how are you confirming pump speed? I'm not a fan of using PWM pump features...why not just set it at 100% or your desired flow and leave it? Is there obvious coolant circulation in your reservoir? It should be very noticeable in a loop like that.= I set the pump to 100% and only saw 2º of difference... Coolant is circulating correctly (I think)

There could be air in the radiator, hard to say, but if the loop is brand new, depending on where the radiator is located and oriented, this might or might not be an issue to check on.

If you shut down the computer and pull the CPU block, is the thermal paste spread evenly with a good contact patch?
= I did not try it due to the lack of the EK thermal paste... But I can say that the thermal paste was applyed with the "all surface cover mode" and I think thermal paste is not the issue...

Did you setup the EK CPU block with the correct jet direction and flow plate? (Does the CPU block in your kit have this option?) No idea... But I took some photos of the loop: IMAGE OF MY LOOP: https://drive.google.com/file/d/1cNs76kNV0j9YuIJhHNNSVgu-tbtrE_KX/view

Detail image: https://drive.google.com/file/d/1Dziw4gsXlFVYbd67Z0LN_e_tOmgKZje4/view?usp=sharing

What are your radiator fans set to at 100% load? =
Can you please explain the question in a different way, I could not underestand it?


What case? How is it installed? Can you post a photo? " = Cooler master mastercase H500M
 
May 12, 2019
32
1
30
"Sounds like lack of flow is not the issue, how are you confirming pump speed? I'm not a fan of using PWM pump features...why not just set it at 100% or your desired flow and leave it? Is there obvious coolant circulation in your reservoir? It should be very noticeable in a loop like that.= I set the pump to 100% and only saw 2º of difference... Coolant is circulating correctly (I think)

There could be air in the radiator, hard to say, but if the loop is brand new, depending on where the radiator is located and oriented, this might or might not be an issue to check on.

If you shut down the computer and pull the CPU block, is the thermal paste spread evenly with a good contact patch?
= I did not try it due to the lack of the EK thermal paste... But I can say that the thermal paste was applyed with the "all surface cover mode" and I think thermal paste is not the issue...
 
May 12, 2019
32
1
30
Did you setup the EK CPU block with the correct jet direction and flow plate? (Does the CPU block in your kit have this option?) No idea... But I took some photos of the loop: IMAGE OF MY LOOP: https://drive.google.com/file/d/1cNs76kNV0j9YuIJhHNNSVgu-tbtrE_KX/view

Detail image: https://drive.google.com/file/d/1Dziw4gsXlFVYbd67Z0LN_e_tOmgKZje4/view?usp=sharing

What are your radiator fans set to at 100% load? =
Can you please explain the question in a different way, I could not underestand it?


What case? How is it installed? Can you post a photo? " = Cooler master mastercase H500M
 

rubix_1011

Contributing Writer
Moderator
Contact patch of the block would only be seen if you pull the block from the CPU to see if it seated equally across the entire CPU IHS. This is typically easy to tell if you have a poor or incorrectly seated block.

What are your radiator fans set to at 100% load? =
Can you please explain the question in a different way, I could not underestand it?
How are you confirming your radiator fans are running at full speed when you are running CPU load testing?

I set the pump to 100% and only saw 2º of difference... Coolant is circulating correctly (I think)

HOW did you set this? If you disconnect the PWM cable and only run on SATA or MOLEX power, the pump should run at 100% speed. Have you attempted this or otherwise validated the pump is running 100% If so, how and what is the pump RPM you observed?

If you shut down the computer and pull the CPU block, is the thermal paste spread evenly with a good contact patch?= I did not try it due to the lack of the EK thermal paste... But I can say that the thermal paste was applyed with the "all surface cover mode" and I think thermal paste is not the issue...

This seems like excessive use of thermal compound and not sure I agree with your dismissal, but I'm not certain how you applied this. Normally it would applied as a single spot about the size of 2 rice grains or small lentil.

Did you setup the EK CPU block with the correct jet direction and flow plate? (Does the CPU block in your kit have this option?) No idea...
Looking at the documentation that came with the kit, it doesn't seem like this is an option, so you might be fine there.




I still reserve that the CPU temps are higher than I would expect unless the CPU is moderately or highly overclocked (is it?)

Is the air coming from the radiator very hot or very warm? Is the radiator itself very hot or very warm? Or is the only indication the temp readings from your software?

Temp dropping immediately when the test is halted may or may not indicate good CPU block contact with the CPU IHS or excessive thermal compound, but does indicate the cooling loop is working.
 
May 12, 2019
32
1
30
Thank you for your response,

"How are you confirming your radiator fans are running at full speed when you are running CPU load testing?" Because I configured them in the bios to be at 100% and also because the noise they produce.

HOW did you set this? If you disconnect the PWM cable and only run on SATA or MOLEX power, the pump should run at 100% speed. Have you attempted this or otherwise validated the pump is running 100% If so, how and what is the pump RPM you observed? I am using a Sata cable. Pump at 100% and no difference in temps.

This seems like excessive use of thermal compound and not sure I agree with your dismissal, but I'm not certain how you applied this. Normally it would applied as a single spot about the size of 2 rice grains or small lentil. I used little thermal paste (single spot about the size of 2 rice grains or small lentil) and I used the rest for other computers I´m building.

I still reserve that the CPU temps are higher than I would expect unless the CPU is moderately or highly overclocked (is it?) The i9-9900K is not overclocked

Is the air coming from the radiator very hot or very warm? Is the radiator itself very hot or very warm? Or is the only indication the temp readings from your software? The radiator (surprisingly ) is not very hot, it is a bit hot but it is cooler than what I expected. The air coming from the radiator is not hot...
I am only measuring the temps with Aida 64.
 

rubix_1011

Contributing Writer
Moderator
OK, so we're evaluating this only from the reported temperatures you are seeing, otherwise, radiator is not excessively hot, nor are there any other indicators of issues. Does your CPU temp get very high doing other things like using applications, watching videos, playing games or benchmarking graphics, memory, network, disk, etc?

What are you using that monitors your CPU temps?

What is your motherboard brand, model and BIOS revision?
 
May 12, 2019
32
1
30
Hi,
The Cpu (When working with word, chrome e.t.c) is between 30º and 35º. This was measured with CPUID HWMonitor.
This tems match with Aida 64 engineer. So I could say that iddle temps are not high.

What are you using that monitors your CPU temps? : Aida 64 engineer, HWMonitor and Speccy.

What is your motherboard brand, model and BIOS revision? : Z390 Aorus master Revision 1Bios F4
 

rubix_1011

Contributing Writer
Moderator
I am not as familiar with Aida64...so I might need to download and try it on my 9700k at home with my full watercooling loop.

Hopefully someone else has some information that can be provided to help bridge any gaps...something just isn't right and I need to figure out what.
 
May 4, 2019
59
6
45
Looks like you might want to rotate the cpu block 90 degrees anti-clockwise to place the outlet above the inlet. That would also help with any potential air bubbles in the water block. Your inlet and outlet are in horizontal alignment, which allows for an air pocket to form above the two fittings.

FWIW, I have a i7-9700K running solid at 5.0GHz (except on the artificial p95 small AVX2-FMA FFT test), with a Corsair 100i Pro. No problem with temperatures up to about 200 watts package power. It's running 8 cores mersenne search at 165 watts and 75 C, with the fans well below top speed.

If you have good thermal contact in the cpu loop you should be able to run 250+ watts on your loop without terrible temperatures.
 
May 12, 2019
32
1
30
Thanks for your answer. I´ll try to rotate it. Also, I´ve got some other info:
-The difference between the "cold" inlet water and the "hot" outlet water if of only 1,2º. So water is not heating to much after going through the CPU. That is why the air comming outside the 360mm radiator is not very hot. The radiator is not hot. Pump is moving coolant and all fans are spining at the correct speed and in the correct way. I upload some images running prime95 V26.6 small FTTs so that you can see all my voltages, powers e.t.c using HWiNFO 64 v6.06-3770 as the hardware monitoring software:

Drive link to the image: https://drive.google.com/file/d/1dPNhyLE0vyTQxQo7kXPocG4ihU8CMBl3/view?usp=sharing


-Possible Solutions that I found:

-"Looks like you might want to rotate the cpu block 90 degrees anti-clockwise to place the outlet above the inlet. That would also help with any potential air bubbles in the water block. Your inlet and outlet are in horizontal alignment, which allows for an air pocket to form above the two fittings. " I´ll try tro do it. Recomendation from mark1776

-Applying some Thermal Grizzly Kryonaut thermal paste correctly

-Sent all info. to the intel, EKWB and Aorus support


Would you recomend me to do something more??
Thanks,
Pablo.
 
May 4, 2019
59
6
45
Frankly, these temperatures don't look too bad for the small fft test, assuming it is running on the AVX2-FMA version. It is not uncommon for a couple of cores to be several degrees warmer than others due to die layout. If the test is not throwing errors, then you are stable. You would like to see somewhat lower temperatures, but this particular test runs hot.

My 9700K runs about the same test/temps at 4.7 GHz clock speed.

Your display does not show the power. Run instead CPUID-HWmonitor, which shows also show the power total for the package. Power is what causes the heat, so it is important to see for cooling analysis.
 

mj.ehsan1999.2nd

Commendable
Jul 22, 2018
16
1
1,515
Hi, I own an i9-9900K and I bought a 360mm radiator with a custom loop ONLY FOR THE CPU (EK Fluid gaming A360G) to cool it.
But I have really high temps at stock clocks: When runing Aida 64for 10 minutes, it throttles in a 5-10%range. With tems really high. CPU at: 83º, CPU core 1: 90ª, CPU c2: 75ª, core 3: 98ª cpu cre 4:
100ª e.t.c

I think these tems are really high for a custom 360mm loop. What do you think? What could be causing the issue???
I think it's a problem with voltages; some motherboards overvoltage CPUs in Auto mode! try lowering the voltage to 1.000volts@stock frequency. also try overclocking to 5GHz@1.32v bcuz I just see some I9-9900k CPUs are on a 1.416v with stock clock speeds . what is your CPU voltage in various conditions ( idle and various stress tests) ? do you use MCE (MultiCore Enhancement )?
 

oczdude8

Distinguished
@oczdude8


Hi,
Thanks for your response.

I have never tried the cpu with other cooler...
CPU temp jumps (when running Aida64 stressing cpu, FPU and cache) very quickly to high temps in about 1-2 seconds.

The fact that the temperature rises almost instantly to a high number implies a large thermal resistance between the CPU die and the water in your cooling loop.

This can mean three things:
-incorrect installation of the CPU water block (poor mounting, bad paste application, wrong inlet/outlet connections etc)
-water bubble in the cpu block, blocking liquid from properly touching copper cold plate
-defective CPU heat spreader (unlikely)
 
May 12, 2019
32
1
30
@mj.ehsan1999.2nd; Hi. MCE is desactivated. Voltages: Idle is of 0.7V. Aida 64 FPU stress test during 2 minutes: Peak voltage o 1.344. MAX.CPU Throttling OF 20%

I´m going to try to reduce the voltage but I am not sure how to do it... Is there a manual of how to reduce voltages in a safe way? (My mother board is a Gigabyte z390 Aorus Master)
Motherboard VRM is heating up to 42º (Which is a lot. Imagine these tems in a motherboard with no heat pipes of the VRM like the Gigabyte one. And Gigabyte VRM´s seems to be the best put there...

I think (as you pointed aout) that there could be a problem with the CPU votages but I´m not sure.

@oczdude8: Thnaks for your response, I contacted with the EK support team and wen´t through these conclusions (I did not tell them the fact about high voltages because I ignored it):
-rotate the cpu block 90 degrees anti-clockwise to place the outlet above the inlet
-Applying some Thermal Grizzly Kryonaut thermal paste correctly
-Using the 360mm rad. and another of 120mm...


I think I will try to do these before applying all those "Extreme solutions removing the block e.t.c:
-Redice the voltage to 1.000 (If I find a safe solution to do it...

WHOLE SYSTEM SPECS:

-Intel core i9 9900K STOCK
-Gigabyte z390 Aorus Master
-EK FLUID GAMING A360G (Whithout cooling the GPU)
-Corsair RM850X
-Cooler mater mastercase H500M
-Nvidia RTX 2070
-Samsung 970EVO plus 500Gb
-Samsung 970 EVO 500Gb

Thanks.
 
May 12, 2019
32
1
30
Hi,
I just undervolt my cpu fromm 1.344-1.380 to 1.212-1.224V when running Aida64 FPU test stock frequencies. Temps has come down from almost all cores hitting 99º to almost all cores staying between 77-89º. Cinebench scores 5CB more and overheating is of 0º. I did not change the loop, thermal paste e.t.c. I did the undervolting using Intel XTU reducing the core voltage ofset to -0,154 V. Is this undervolting OK?

Also, Would you recommend me to change the thermal paste from the EK one to the thermal grizzly kryonaut that I have?
Would you recommend me to rotate the cpu block 90 degrees anti-clockwise to place the outlet above the inlet ?
And finally, would I obtain better results by adding the other 120mm RAD that I have.

Although temps are cooler, they still high...
Thanks.