Warm coolant temps with 3 360 rads?

[DarkeVortex]

Hey everyone, here’s my setup:

 

relevant specs—

cpu: Ryzen 9 5950X

gpu: 3090 Strix OC

motherboard: Crosshair VIII dark hero

case: O11-D XL (yep, basic I know!)

 

Cooling—

EK Crosshair VIII hero monoblock

EK 3090 Strix full cover block

EK 3090 Strix active backplate

2x EK 360 PE rads

1x EK 360 SE rad

9x LianLi UniFan AL120 (top/bottom are intake, pull, and side is exhaust, push)

 

after running OCCT on power mode for 30 min, I get the following:

CPU=67C

GPU=60C

water=43C

ambient=28C.

this is with fans and pump maxed.

 

from what I have read online, these coolant temps are getting pretty high up there. When gaming in light games (like rocket league) for ~2h, it sometimes crosses 45C.

 

anyone know what could be causing high coolant temps?

[SavageNeo]

Just now, DarkeVortex said:

anyone know what could be causing high coolant temps?

probably the strix 3090 oc. 

[For Science!]

4 minutes ago, DarkeVortex said:

Hey everyone, here’s my setup:

 

relevant specs—

cpu: Ryzen 9 5950X

gpu: 3090 Strix OC

motherboard: Crosshair VIII dark hero

case: O11-D XL (yep, basic I know!)

 

Cooling—

EK Crosshair VIII hero monoblock

EK 3090 Strix full cover block

EK 3090 Strix active backplate

2x EK 360 PE rads

1x EK 360 SE rad

9x LianLi UniFan AL120 (top/bottom are intake, pull, and side is exhaust, push)

 

after running OCCT on power mode for 30 min, I get the following:

CPU=67C

GPU=60C

water=43C

ambient=28C.

this is with fans and pump maxed.

 

from what I have read online, these coolant temps are getting pretty high up there. When gaming in light games (like rocket league) for ~2h, it sometimes crosses 45C.

 

anyone know what could be causing high coolant temps?

VRM heat being offloaded into the loop. I don't think anything is wrong with the loop, perhaps weak fans. The bottom line is its a lot of heat to take out with the 3090, and the CPU VRMs

[SpookyCitrus]

With those specs I'd say the coolant and hardware temps under load are more than fine. That's a lot of heat to dissipate the only thing I can think of to improve the coolant temps would be to get some more airflow going through the rads. Maybe add some extra fans for push/pull to increase airflow, other than that I don't see anything else you really can do besides going with a whole new case/cooling loop.

[AnonymousGuy]

I have a shunted 1080Ti that consumes similar power to a 3090 stock and it works fine with a 240mm radiator with some Silent Wings 3 push pull.  It gets to 45-50C water but it still works.

 

I feel like 4x the radiator space should be doing a lot better unless the fans really aren't good.

[fonzz1e]

if u remove all the dust filters does the coolant temp improve?

how about bottom & side intake while top exhaust?

[airborne spoon]

seems about right that is a lot of heat you are trying to dissipate.

perhaps try 1 intake and 2 exhaust instead of 2 intake and 1 exhaust you're blowing a lot of hot air into the computer that way, yes you're pulling the coldest air in but it might be something to try.

[DarkeVortex]

10 hours ago, fonzz1e said:

if u remove all the dust filters does the coolant temp improve?

how about bottom & side intake while top exhaust?

Filters are off, and it wasn't too much worse with the filters on.

 

I tried bottom and side intake with top exhaust and didn't see much change.

[DarkeVortex]

3 hours ago, airborne spoon said:

seems about right that is a lot of heat you are trying to dissipate.

perhaps try 1 intake and 2 exhaust instead of 2 intake and 1 exhaust you're blowing a lot of hot air into the computer that way, yes you're pulling the coldest air in but it might be something to try.

I could give that a try, good to test at least

[DarkeVortex]

On 8/30/2021 at 11:12 AM, For Science! said:

VRM heat being offloaded into the loop. I don't think anything is wrong with the loop, perhaps weak fans. The bottom line is its a lot of heat to take out with the 3090, and the CPU VRMs

So you think if i switched off a monoblock to a regular CPU block it would get better? Would I see any performance issues not watercooling the VRMs?

[DarkeVortex]

On 8/30/2021 at 11:08 AM, SavageNeo said:

probably the strix 3090 oc. 

That's fair. I just thought with 3 360 rads it would be enough, but maybe not.

[DarkeVortex]

On 8/30/2021 at 2:27 PM, AnonymousGuy said:

I have a shunted 1080Ti that consumes similar power to a 3090 stock and it works fine with a 240mm radiator with some Silent Wings 3 push pull.  It gets to 45-50C water but it still works.

 

I feel like 4x the radiator space should be doing a lot better unless the fans really aren't good.

I would think so too :/

[airborne spoon]

9 minutes ago, DarkeVortex said:

I could give that a try, good to test at least

testing different things is always helpful to see if there is any tangible gains to be had.

9 minutes ago, DarkeVortex said:

So you think if i switched off a monoblock to a regular CPU block it would get better? Would I see any performance issues not watercooling the VRMs?

Technically yes because there is less heat going into the loop. But with watercooled VRM's you get better performance and a little extra heat going into a loop like that should be negligible.

 

Also FYSA instead of making 5 different comments you can just hit the multi-quote button and then reply to multiple people in one single comment like I have done above. It makes for a cleaner read of the thread.

[For Science!]

18 minutes ago, DarkeVortex said:

So you think if i switched off a monoblock to a regular CPU block it would get better? Would I see any performance issues not watercooling the VRMs?

Yes the load on the loop will be reduced significantly. Unless you are doing extreme OCing, heatsinked VRMs air cooled will be more than plenty

[fonzz1e]

5 hours ago, DarkeVortex said:

That's fair. I just thought with 3 360 rads it would be enough, but maybe not.

what's your fan rpm during your stress test?

[DarkeVortex]

13 hours ago, fonzz1e said:

what's your fan rpm during your stress test?

1900, max on these fans.

[DarkeVortex]

18 hours ago, For Science! said:

Yes the load on the loop will be reduced significantly. Unless you are doing extreme OCing, heatsinked VRMs air cooled will be more than plenty

Gotcha. Might swap over then, though I still think the monoblock looks so good...

[DarkeVortex]

18 hours ago, airborne spoon said:

testing different things is always helpful to see if there is any tangible gains to be had.

Technically yes because there is less heat going into the loop. But with watercooled VRM's you get better performance and a little extra heat going into a loop like that should be negligible.

 

Also FYSA instead of making 5 different comments you can just hit the multi-quote button and then reply to multiple people in one single comment like I have done above. It makes for a cleaner read of the thread.

Gotcha. So are these temperatures not like absurd then, or do I still need to worry about it?

and thanks for pointing that out! ...I read it about 5sec too late though, but now I know for next time!

[Curufinwe_wins]

19 hours ago, For Science! said:

Yes the load on the loop will be reduced significantly. Unless you are doing extreme OCing, heatsinked VRMs air cooled will be more than plenty

This is not true. The total load from VRM is marginal compared to the rest of the system.

 

 

Anyways @OP This is not an unreasonable synthetic full load. 60s for both components is not awful and 40s on the coolant is on the higher side, but it's also not unreasonable.

 

Two things to try to understand depending on your situation. A is airflow, B is coolant flow. Keeping A constant while changing B should tell you at what speed your pump needs to run to make moving the heat not a primary bottleneck (more a CPU issue than a GPU one generally, and even with a complex load 70% on a D5 or DDC tends to be enough).

 

Separately, you can also investigate how changing individual fan speeds on each rad makes a difference while keeping the pump speed constant.

 

The other factor to consider is your ambient temp. I'm not going to try to pretend that 28C is unreasonably warm, but if you compare to people who keep their ambient to say 23C, all of your temps are about 5C higher than what they would have. Do that 5C offset and now you are talking about high 30s coolants and high 50s low 60s component, which isn't bad at all.

[For Science!]

6 minutes ago, Curufinwe_wins said:

This is not true. The total load from VRM is marginal compared to the rest of the system.

 

 

Anyways @OP This is not an unreasonable synthetic full load. 60s for both components is not awful and 40s on the coolant is on the higher side, but it's also not unreasonable.

 

Two things to try to understand depending on your situation. A is airflow, B is coolant flow. Keeping A constant while changing B should tell you at what speed your pump needs to run to make moving the heat not a primary bottleneck (more a CPU issue than a GPU one generally, and even with a complex load 70% on a D5 or DDC tends to be enough).

 

Separately, you can also investigate how changing individual fan speeds on each rad makes a difference while keeping the pump speed constant.

 

The other factor to consider is your ambient temp. I'm not going to try to pretend that 28C is unreasonably warm, but if you compare to people who keep their ambient to say 23C, all of your temps are about 5C higher than what they would have. Do that 5C offset and now you are talking about high 30s coolants and high 50s low 60s component, which isn't bad at all.

VRMs produce a tonne of heat, especially in an self-OCing system like Ryzen. They are rated to be operational even at 105 degrees and watercooling reduces this dramatically, at the cost of the thermal capacity of the loop. 

 

On the otherhand coolant flow minimally impacts the performance of the loop and is not a factor worth ramping up and down depending on the load. You can have very minimal flow and the temps will only differ by 2-3 degrees

[Hairless Monkey Boy]

1 hour ago, For Science! said:

VRMs produce a tonne of heat, especially in an self-OCing system like Ryzen. They are rated to be operational even at 105 degrees and watercooling reduces this dramatically, at the cost of the thermal capacity of the loop. 

The VRMs get hot for sure, but that is not because they are producing a ton of heat, it's because the heat they produce is pretty dense and the stock heatsinks on motherboards are usually not very good (and they don't really have to be tbh). If we assume a low efficiency of 80%, then they are producing 20% of the heat of the CPU, so that's around 30W with a 5950X at 150W. It's significant, but not game changing.

 

@DarkeVortexAs @Curufinwe_winssaid, I think the primary issue here is the ambient temp. If we adjust it down to 22C, which is what mine usually is, then your coolant temp would be ~37°C max, which is what mine usually is with an OCed 9900K and RTX 2080 on 6X 140mm worth of radiator. So I think you are fine.

[Curufinwe_wins]

1 hour ago, For Science! said:

VRMs produce a tonne of heat, especially in an self-OCing system like Ryzen. They are rated to be operational even at 105 degrees and watercooling reduces this dramatically, at the cost of the thermal capacity of the loop. 

 

On the otherhand coolant flow minimally impacts the performance of the loop and is not a factor worth ramping up and down depending on the load. You can have very minimal flow and the temps will only differ by 2-3 degrees

As @HairlessMonkeyBoysays. They do not produce a ton of heat. Temperature =/= heat which is self evident that a huge watercooling loop functions to keep a 350w gpu + 100-300W cpu down to 60C but tiny VRM heatsinks and ambient non-targeted airflow is enough to keep them operating safely, even if at 105C. Even if the VRMs consume 100W (which is not happening, basically ever outside of LN2 and similar and maybe not even then), compared to the 600W the rest of the system loop is producing, that is only a 20% difference. VRM power draws of 10-30W is much more reasonable. They are generally very efficient (think closer to 90% than the 80% referenced above). Which becomes basically margin of error in the whole loop.

 

As far as coolant flow goes. It matters like crazy until it basically doesn't matter at all. I don't believe it is his problem (I literally stated 70% on most D5 and DDC pumps and normal loops is more than enough and he claimed to be running it at 100%). I just told him how to go about checking.

 

21 minutes ago, HairlessMonkeyBoy said:

The VRMs get hot for sure, but that is not because they are producing a ton of heat, it's because the heat they produce is pretty dense and the stock heatsinks on motherboards are usually not very good (and they don't really have to be tbh). If we assume a low efficiency of 80%, then they are producing 20% of the heat of the CPU, so that's around 30W with a 5950X at 150W. It's significant, but not game changing.

 

@DarkeVortexAs @Curufinwe_winssaid, I think the primary issue here is the ambient temp. If we adjust it down to 22C, which is what mine usually is, then your coolant temp would be ~37°C max, which is what mine usually is with an OCed 9900K and RTX 2080 on 6X 140mm worth of radiator. So I think you are fine.

Well put. I do think there is potential that a number of the fans are not doing well with the rather poor airflow of the O11d, especially if he is on a carpeted surface, but overall his temps are not *that* bad, so my perspective is go ahead and play with controlled testing, but don't worry that much either.

[For Science!]

36 minutes ago, HairlessMonkeyBoy said:

The VRMs get hot for sure, but that is not because they are producing a ton of heat, it's because the heat they produce is pretty dense and the stock heatsinks on motherboards are usually not very good (and they don't really have to be tbh). If we assume a low efficiency of 80%, then they are producing 20% of the heat of the CPU, so that's around 30W with a 5950X at 150W. It's significant, but not game changing.

 

@DarkeVortexAs @Curufinwe_winssaid, I think the primary issue here is the ambient temp. If we adjust it down to 22C, which is what mine usually is, then your coolant temp would be ~37°C max, which is what mine usually is with an OCed 9900K and RTX 2080 on 6X 140mm worth of radiator. So I think you are fine.

 

16 minutes ago, Curufinwe_wins said:

As @HairlessMonkeyBoysays. They do not produce a ton of heat. Temperature =/= heat which is self evident that a huge watercooling loop functions to keep a 350w gpu + 100-300W cpu down to 60C but tiny VRM heatsinks and ambient non-targeted airflow is enough to keep them operating safely, even if at 105C. Even if the VRMs consume 100W (which is not happening, basically ever outside of LN2 and similar and maybe not even then), compared to the 600W the rest of the system loop is producing, that is only a 20% difference. VRM power draws of 10-30W is much more reasonable. They are generally very efficient (think closer to 90% than the 80% referenced above). Which becomes basically margin of error in the whole loop.

 

As far as coolant flow goes. It matters like crazy until it basically doesn't matter at all. I don't believe it is his problem (I literally stated 70% on most D5 and DDC pumps and normal loops is more than enough and he claimed to be running it at 100%). I just told him how to go about checking.

 

Well put. I do think there is potential that a number of the fans are not doing well with the rather poor airflow of the O11d, especially if he is on a carpeted surface, but overall his temps are not *that* bad, so my perspective is go ahead and play with controlled testing, but don't worry that much either.

Okay, your points have been noted, but I would like to see the experiment performed by OP and the results to be reported. Certainly for me, when I went from a Monoblocked Z270 board for a 7700K at a moderate OC of 4.8/4.9 GHz to a regular CPU waterblock, I had a notable reduction in coolant temperature.