Watercooling setup performing as expected?

[Stian Kalstad]

Hi,

 

I have set up a 12900K and 3080TI with watercooling parts from EK; EK-Quantum Velocity² for the CPU and EK-Quantum Vector² FTW3 RTX 3080/90 ABP Set for the GPU. The loop is set up with two 360mm radiators, six 120mm fans and pump-reservoir combo (all parts from EK). This is my first experience with watercooling.

 

I have run some stress tests and benchmarks and I was a little surprised to see that some of the cores in the CPU were touching thermal-throttle temps when running torture test in prime95 in stock speeds. I thought this loop would would be more than enough to avoid temperatures in the high 90`s? Below is the readings of core clocks and core temps from HWinfo when running Prime95 and Cinebench r23.

 

Does the high temperatures while running Prime95 indicate some kind of contact problems between the chip and the waterblock? Or should these temperatures be expected when running this test? The temperatures when running Cinebench seems fine (score 26744). The temperatures when running stability test in AIDA64 also seems fine (below 80 degrees). I have already tried to reapply the thermal paste once without any change, and the screws for the waterblock are tightened with a torque-screwdriver so the contact pressure should be OK.

 

Thanks in advice,

 

Stian

 

 

 

[LIGISTX]

7 minutes ago, Stian Kalstad said:

Hi,

 

I have set up a 12900K and 3080TI with watercooling parts from EK; EK-Quantum Velocity² for the CPU and EK-Quantum Vector² FTW3 RTX 3080/90 ABP Set for the GPU. The loop is set up with two 360mm radiators, six 120mm fans and pump-reservoir combo (all parts from EK). This is my first experience with watercooling.

 

I have run some stress tests and benchmarks and I was a little surprised to see that some of the cores in the CPU were touching thermal-throttle temps when running torture test in prime95 in stock speeds. I thought this loop would would be more than enough to avoid temperatures in the high 90`s? Below is the readings of core clocks and core temps from HWinfo when running Prime95 and Cinebench r23.

 

Does the high temperatures while running Prime95 indicate some kind of contact problems between the chip and the waterblock? Or should these temperatures be expected when running this test? The temperatures when running Cinebench seems fine (score 26744). The temperatures when running stability test in AIDA64 also seems fine (below 80 degrees). I have already tried to reapply the thermal paste once without any change, and the screws for the waterblock are tightened with a torque-screwdriver so the contact pressure should be OK.

 

Thanks in advice,

 

Stian

 

 

 

Use something other then P95- it’s more of a power virus then a useful test these days. 
 

Use AIDA64, see what it gets the temps to. I would bet it will drop 15c. 

[freeagent]

25 minutes ago, LIGISTX said:

Use AIDA64, see what it gets the temps to. I would bet it will drop 15c. 

Aida has spiced things up a bit since I last ran it with an old version..

[Hairless Monkey]

26 minutes ago, LIGISTX said:

Use AIDA64

34 minutes ago, Stian Kalstad said:

The temperatures when running stability test in AIDA64 also seems fine (below 80 degrees).

...

[Stian Kalstad]

55 minutes ago, LIGISTX said:

Use something other then P95- it’s more of a power virus then a useful test these days. 
 

Use AIDA64, see what it gets the temps to. I would bet it will drop 15c. 

The FPU stress test ran a little bit hotter than I said. Around 85c after 40 min. Are these temps within what should be expected with this cooling-setup?

 

[LIGISTX]

33 minutes ago, Stian Kalstad said:

The FPU stress test ran a little bit hotter than I said. Around 85c after 40 min. Are these temps within what should be expected with this cooling-setup?

 

Seems a bit high, but it’s possible your chip is just a warm one. What thermal paste are you using?

[Stian Kalstad]

17 minutes ago, LIGISTX said:

Seems a bit high, but it’s possible your chip is just a warm one. What thermal paste are you using?

Not much room for overclocking, which is kind of dissapointing. Thermal grizzly, came along with the waterblock. Thats a decent paste, isn’t it?

[LIGISTX]

25 minutes ago, Stian Kalstad said:

Not much room for overclocking, which is kind of dissapointing. Thermal grizzly, came along with the waterblock. Thats a decent paste, isn’t it?

They have a few different versions, but the one EK uses is a good one, ya.

 

OCing on new Intel hardware is hard, it’s already pushed pretty much to the max. You can likely push it a little more, but If your already at 85 in Aida there isn’t much headroom. I had success sanding down (lapping) my EK block for use with my 10700k, it dropped a noticeable amount of temp, but you may have gotten a not so great 10900k. My 9900k was similar, it was clearly a chip that just made the cut…. It ran HOT. 

[fonzz1e]

p95 is for stress testing (aka how much can your loop handle) 

 

@Stian Kalstadduring your p95 runs, did u log the coolant temp? if it didn't change much it might be contact between cpu & block

 

I recall thermal grizzly did release a mounting plate for alder lake, maybe try get it

 

EDIT: found it https://www.thermal-grizzly.com/en/products/523-cpu-contact-frame-for-12th-gen-intel-by-der8auer-en

 

[Stian Kalstad]

4 hours ago, fonzz1e said:

p95 is for stress testing (aka how much can your loop handle) 

 

@Stian Kalstadduring your p95 runs, did u log the coolant temp? if it didn't change much it might be contact between cpu & block

 

I recall thermal grizzly did release a mounting plate for alder lake, maybe try get it

 

EDIT: found it https://www.thermal-grizzly.com/en/products/523-cpu-contact-frame-for-12th-gen-intel-by-der8auer-en

 

Had a temperature sensor from a commander pro stuck in one of the radiator-fins, the temperature increased about 3c during the p95 test. I could not feel any difference when touching the reservoir or the tubes. When running furmark to stress the GPU, I could really feel the temperature-difference of the coolant just by touching the tubes. I have read about the issue with bend CPUs with the LGA1700 socket, but I didn't know about this contact frame, thanks. I have ordered that along with some kryonaut paste.

 

If that doesn't work I will consider trying to lapping down the waterblock.

Edited June 15, 2022 by Stian Kalstad

[jaslion]

What is your case and how is the loop set up? All we know so far is 2 360 rads and 6 fans rn. A picture of the setup would be good too.

[Stian Kalstad]

2 hours ago, jaslion said:

What is your case and how is the loop set up? All we know so far is 2 360 rads and 6 fans rn. A picture of the setup would be good too.

It’s a lian-li O11D. The side panels have never been on, so the airflow should not be a issue at this point;) The rads are installed at the bottom and top of the case, with pull-config on the fans (i.e. mounted on the top of the rads).

 

The loop is set up as following:

From the outlet of the pump/reservoir-combo -> bottom rad -> top rad -> GPU block -> CPU block -> back into the inlet of the pump/res-combo. 

 

The pump is a D5 (EK-Quantum Kinetic FLT 360 D5 PWM D-RGB - Plexi).

 

Have used soft tubing 10/16 mm and EK-CryoFuel as coolant. Will post a picture of the setup in the evening:)

[jaslion]

16 minutes ago, Stian Kalstad said:

It’s a lian-li O11D. The side panels have never been on, so the airflow should not be a issue at this point;) The rads are installed at the bottom and top of the case, with pull-config on the fans (i.e. mounted on the top of the rads).

 

The loop is set up as following:

From the outlet of the pump/reservoir-combo -> bottom rad -> top rad -> GPU block -> CPU block -> back into the inlet of the pump/res-combo. 

 

The pump is a D5 (EK-Quantum Kinetic FLT 360 D5 PWM D-RGB - Plexi).

 

Have used soft tubing 10/16 mm and EK-CryoFuel as coolant. Will post a picture of the setup in the evening:)

These fans are made to push air not pull they are worse at that. Id honestly do the tried and true all fans intake (with the side panels on to have proper directional airflow) and nothing in the back config.

[LIGISTX]

3 hours ago, Stian Kalstad said:

From the outlet of the pump/reservoir-combo -> bottom rad -> top rad -> GPU block -> CPU block -> back into the inlet of the pump/res-combo

It likely isn’t a big deal, but ideally you want your exhaust radiator first in the loop so you shed some heat before you pump it back into the case via the intake rad. Having the intake rad first creates a positive feedback loop where it heats the air up that is used to cool the exhaust rad. Positive feedback loop is not really true, since it isn’t a runaway affect, but it does have an impact, but not a huge one. 
 

3 hours ago, jaslion said:

These fans are made to push air not pull they are worse at that.

Fans really shouldn’t care much about push or pull. Some do better at creating static pressure, some do worse, but the pressure differential will be the same in push or pull. The air it’s pushing has to come from somewhere…. 

[Stian Kalstad]

1 hour ago, LIGISTX said:

It likely isn’t a big deal, but ideally you want your exhaust radiator first in the loop so you shed some heat before you pump it back into the case via the intake rad. Having the intake rad first creates a positive feedback loop where it heats the air up that is used to cool the exhaust rad. Positive feedback loop is not really true, since it isn’t a runaway affect, but it does have an impact, but not a huge one. 

I see the point of having the exhaust before the intake in the loop. But as the coolant doesn’t really get any hotter during the CPU stress test, I dont think thats the main problem.

 

@jaslion picture of the setup is attached. The thought behind having the fans in pull-config was that it would be easier to remove dust from the rads. Are there any obvious flaws in the setup? (Except for the horrible tubing, will fix that later when the build and troubleshooting is complete;)) 

If there isn’t any major errors in the setup of the loop, I think the measures that would most likely have the biggest impact on the cpu-temps would be to try the Thermal-Grizzly contact frame and lapping down the waterblock, agreed? Or i could also be that the scilicon in my chip just isn’t that good:|

 

[jaslion]

2 hours ago, LIGISTX said:

It likely isn’t a big deal, but ideally you want your exhaust radiator first in the loop so you shed some heat before you pump it back into the case via the intake rad. Having the intake rad first creates a positive feedback loop where it heats the air up that is used to cool the exhaust rad. Positive feedback loop is not really true, since it isn’t a runaway affect, but it does have an impact, but not a huge one. 
 

Fans really shouldn’t care much about push or pull. Some do better at creating static pressure, some do worse, but the pressure differential will be the same in push or pull. The air it’s pushing has to come from somewhere…. 

It really depends. I have alphacool fans that kinda really suck in pull config due to cuts in the blade at the end. I also thought it wouldn't matter but it did sadly.

[jaslion]

4 minutes ago, Stian Kalstad said:

I see the point of having the exhaust before the intake in the loop. But as the coolant doesn’t really get any hotter during the CPU stress test, I dont think thats the main problem.

 

@jaslion picture of the setup is attached. The thought behind having the fans in pull-config was that it would be easier to remove dust from the rads. Are there any obvious flaws in the setup? (Except for the horrible tubing, will fix that later when the build and troubleshooting is complete;)) 

If there isn’t any major errors in the setup of the loop, I think the measures that would most likely have the biggest impact on the cpu-temps would be to try the Thermal-Grizzly contact frame and lapping down the waterblock, agreed? Or i could also be that the scilicon in my chip just isn’t that good:|

 

Is your pump at 100% speed?

 

Is that a kink in the tube that goes from the buttom rad to the cpu?