12700k AIO cooling

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
That is very waterblock and radiator specific though so you need the exact same components for the flow estimate to be valid. The change in temps depends a lot on the water block. Some waterblocks see changes of several degrees while others are measured in decimals between 0.5GPM and 1GPM flow rate.

The radiator setup has much more impact than flow rate IMO. Setting up all radiators as either intake or outtake will have more impact. Some radiators as intake and some removing air in the same loop means you get very little cooling from the ones removing air from the case unless you are moving lots of air through the case (additional intake fans). The air coming out of a slim radiator will be closer to water temp than ambient and a fat radiator will make the air close to water temp. The radiators drawing air from the case will operate at much lower efficiency due to the air being much hotter. If all of your fans are attached to radiators then I would expect a 480 as intake to be close to a 360 as intake with 360+120 as outtake.
Of course it is.
For data to be any good it has to be specific to the equipment used in that series of tests.

All water cooled systems have a water loop composed of waterblock, pump and radiator/s plus airflow through radiators. That's 2 complete systems; water loop and airflow. Loop must be a balanced group of components and airflow much supply needed cool air. Both then work together to give good heat transfer from component to coolant to airflow..

You can have all the radiators in the world but if heat isn't moving from waterblock to coolant your temps will be high.
If airflow isn't removing heat from coolant in radiators you will have high temps.

All must work together for things to be cool. ;)
 
Joined
Apr 9, 2021
Messages
915
Here's a link to Martin's Liquid Lab about number of waterblocks and flowrate to temperature change. Obviously a very old article but still applicable. 0.5GPM is 113.6L/h. You can see increasing flow rate from 0.5GPM to 1.0GPM lowered temp 2.06c while from 1.0 to 1.5 only lowered temp 0.75c, and each additional 0.5GPM increase in flow game less and less temp decrease.
https://www.techpowerup.com/MartinsLiquidLab/MartinsFlowRateEstimator.html

Quick test with my EVGA CLC-280 and 5960X at various pump speeds. Each load level was allowed to run for 10 minutes in order to reach steady state, and I gave the system 2 minutes to stabilize after each pump speed change. Temps reported are package temp averaged over 1 minute. Ambient temp measured 6 inches from the intake case fan was 24.0C +/- 0.1C for the duration. The fans were on the curve I use daily, which resulted in 100% for the 200W and 280W load tests and significantly less at 140W.

AIDA64 Stress Test, CPU only
140W load, 2800RPM: 51.5C
140W load, 2300RPM: 51.8C
140W load, 1800RPM: 52.3C

Prime 95 26.6 Small FFT
200W load, 2800RPM: 61.2C
200W load, 2300RPM: 61.8C
200W load, 1800RPM: 63.0C

Prime 95 30.4 Small FFT
260W load, 2800RPM: 72.6C
260W load, 2300RPM: 73.6C
260W load, 1800RPM: 75.3C

One thing that I observed but did not record this time around is that while the averages are not that far apart, the spikes were much higher at the minimum pump speed. In the 260W test, with the highest pump speed, the highest temperature was 74C, but with the slowest pump speed it was 78C, which is more than the difference between the averages.

Judging by the fact that the temperatures keep dropping at a decent clip all the way to the maximum speed, I would imagine that more flow would lead to a further reduction in temperature. I wonder how low the temperatures would eventually get if I were running a D5 pump? I stuck with the AIO on this machine because the case only has room for a single 280mm radiator, so I didn't see the point of going custom, but maybe there would be a significant improvement. Might do a CPU only loop and see what happens.
 
Last edited:

evhvis

Limp Gawd
Joined
Feb 12, 2021
Messages
326
Quick test with my EVGA CLC-280 and 5960X at various pump speeds. Each load level was allowed to run for 10 minutes in order to reach steady state, and I gave the system 2 minutes to stabilize after each pump speed change. Temps reported are package temp averaged over 1 minute. Ambient temp measured 6 inches from the intake case fan was 24.0C +/- 0.1C for the duration. The fans were on the curve I use daily, which resulted in 100% for the 200W and 280W load tests and significantly less at 140W.

AIDA64 Stress Test, CPU only
140W load, 2800RPM: 51.5C
140W load, 2300RPM: 51.8C
140W load, 1800RPM: 52.3C

Prime 95 26.6 Small FFT
200W load, 2800RPM: 61.2C
200W load, 2300RPM: 61.8C
200W load, 1800RPM: 63.0C

Prime 95 30.4 Small FFT
260W load, 2800RPM: 72.6C
260W load, 2300RPM: 73.6C
260W load, 1800RPM: 75.3C

One thing that I observed but did not record this time around is that while the averages are not that far apart, the spikes were much higher at the minimum pump speed. In the 260W test, with the highest pump speed, the highest temperature was 74C, but with the slowest pump speed it was 78C, which is more than the difference between the averages.

Judging by the fact that the temperatures keep dropping at a decent clip all the way to the maximum speed, I would imagine that more flow would lead to a further reduction in temperature. I wonder how low the temperatures would eventually get if I were running a D5 pump? I stuck with the AIO on this machine because the case only has room for a single 280mm radiator, so I didn't see the point of going custom, but maybe there would be a significant improvement. Might do a CPU only loop and see what happens.
The big difference in max temps is due to AIO pumps having fairly low flow rates so you often need maximum speed to get good cooling effect, but it then ends up as a dilemma between noise vs cooling so a lot settle on medium speed on the pump. The major advantages a custom loop has, outside of being able to run bigger and better radiators, are that the CPU blocks are generally of much higher quality than what you will find in an AIO and flow rate. Drawbacks with custom loops are size, cost and complexity. For most an AIO will be good enough though.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
Quick test with my EVGA CLC-280 and 5960X at various pump speeds. Each load level was allowed to run for 10 minutes in order to reach steady state, and I gave the system 2 minutes to stabilize after each pump speed change. Temps reported are package temp averaged over 1 minute. Ambient temp measured 6 inches from the intake case fan was 24.0C +/- 0.1C for the duration. The fans were on the curve I use daily, which resulted in 100% for the 200W and 280W load tests and significantly less at 140W.

AIDA64 Stress Test, CPU only
140W load, 2800RPM: 51.5C
140W load, 2300RPM: 51.8C
140W load, 1800RPM: 52.3C

Prime 95 26.6 Small FFT
200W load, 2800RPM: 61.2C
200W load, 2300RPM: 61.8C
200W load, 1800RPM: 63.0C

Prime 95 30.4 Small FFT
260W load, 2800RPM: 72.6C
260W load, 2300RPM: 73.6C
260W load, 1800RPM: 75.3C

One thing that I observed but did not record this time around is that while the averages are not that far apart, the spikes were much higher at the minimum pump speed. In the 260W test, with the highest pump speed, the highest temperature was 74C, but with the slowest pump speed it was 78C, which is more than the difference between the averages.

Judging by the fact that the temperatures keep dropping at a decent clip all the way to the maximum speed, I would imagine that more flow would lead to a further reduction in temperature. I wonder how low the temperatures would eventually get if I were running a D5 pump? I stuck with the AIO on this machine because the case only has room for a single 280mm radiator, so I didn't see the point of going custom, but maybe there would be a significant improvement. Might do a CPU only loop and see what happens.
Makes sense for spikes to be higher. Lower coolant flow rate means it's getting hotter while in water block so can't absorb heat as fast .. end result is spiking heat isn't drawn away as fans so we see high spike temps. Honestly as long as max temps are below whatever your CPU starts throttling it doesn't matter. That said I like to keep temps below 85-90c peaks with average about 70-75c .. that's at maximum function load. ;)

The big difference in max temps is due to AIO pumps having fairly low flow rates so you often need maximum speed to get good cooling effect, but it then ends up as a dilemma between noise vs cooling so a lot settle on medium speed on the pump. The major advantages a custom loop has, outside of being able to run bigger and better radiators, are that the CPU blocks are generally of much higher quality than what you will find in an AIO and flow rate. Drawbacks with custom loops are size, cost and complexity. For most an AIO will be good enough though.
I believe JSHamlet234 has custom loop w/ 840mm of rad on EK-D5 pump rated up to 1500L/h, so even at 20-30% speed it's flowing several times more coolant than CLCs flow.
Wrong user, JSHamlet234 should be xDiVolatilX. Sorry about that.

 
Last edited:

evhvis

Limp Gawd
Joined
Feb 12, 2021
Messages
326
I believe JSHamlet234 has custom loop w/ 840mm of rad on EK-D5 pump rated up to 1500L/h, so even at 20-30% speed it's flowing several times more coolant than CLCs flow.
He wrote EVGA CLC-280 in the post I quoted.

A D5 at 20% with 840mm of rad will be between low and almost no flow depending on the restriction of the rads and blocks used. If they are high restriction radiators and blocks then you might be well below CLC flows at 20% and possibly max out at 1.5-2.5l/m at 100% pump speed. The D5s lacks headpressure, but are good at moving liquid quietly in moderately restrictive loops. If it is a high restriction loop then a DDC is usually better due to much higher headpressure.
 

xDiVolatilX

Limp Gawd
Joined
Jul 24, 2021
Messages
385
That is very waterblock and radiator specific though so you need the exact same components for the flow estimate to be valid. The change in temps depends a lot on the water block. Some waterblocks see changes of several degrees while others are measured in decimals between 0.5GPM and 1GPM flow rate.

The radiator setup has much more impact than flow rate IMO. Setting up all radiators as either intake or outtake will have more impact. Some radiators as intake and some removing air in the same loop means you get very little cooling from the ones removing air from the case unless you are moving lots of air through the case (additional intake fans). The air coming out of a slim radiator will be closer to water temp than ambient and a fat radiator will make the air close to water temp. The radiators drawing air from the case will operate at much lower efficiency due to the air being much hotter. If all of your fans are attached to radiators then I would expect a 480 as intake to be close to a 360 as intake with 360+120 as outtake.

All 840mm of my rads are exhaust. I have modified a corsair graphite series 780T full tower with a half open side panel housing a 200mm coolermaster side intake fan blasting the whole board with fresh cold air as well as a bottom 140mm fractal fan intaking from below. This way I have all 3 rads as exhaust and the inside of the case is cold as all hot air is exhausted. I found this method to have the best thermals with lowest fan speeds.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
He wrote EVGA CLC-280 in the post I quoted.

A D5 at 20% with 840mm of rad will be between low and almost no flow depending on the restriction of the rads and blocks used. If they are high restriction radiators and blocks then you might be well below CLC flows at 20% and possibly max out at 1.5-2.5l/m at 100% pump speed. The D5s lacks headpressure, but are good at moving liquid quietly in moderately restrictive loops. If it is a high restriction loop then a DDC is usually better due to much higher headpressure.
That's an assumption I don't agree with you, but we are both making assumptions. I seriously doubt 840mm of rad and 2 waterblocks will limit D5 to only 1.5-2.5l/m is correct that's 90L/h to 150L/h which is significantly than most CLCs. I know specs of. D5 have way more heat pressure than any other pump at 3.9m and 1500L/h and DDC is only slightly more at 4.2m but only rated to flow 1000L/h which is 66% of what D5 flows. To my way of thinking slightly higher flowrate with little lower lift balances out against slightly less flowrate and little more lift rating pretty much balances their performance to be about equal. While majority of CLCs do not publish pump specs, the few that do are like H50 (Asetek) is only 40L/h with no lift spec, Asetek 545LC is 61L/h and Cool IT Eco III 34.2L/h. Alphacool AIO pumps are rated 72L/h or 0.6m and 100L/h or 1m lift depending model. ID Cooling CLC pumps are all rated 96-106L/h but one which is rated 1.2-1.3m lift with their Dashflow being 450L/h or 3m lift.

xDiVolatilX system is cooling just fine, so obviously 20-30% pump speed is enough.​

 

evhvis

Limp Gawd
Joined
Feb 12, 2021
Messages
326
That's an assumption I don't agree with you, but we are both making assumptions. I seriously doubt 840mm of rad and 2 waterblocks will limit D5 to only 1.5-2.5l/m is correct that's 90L/h to 150L/h which is significantly than most CLCs. I know specs of. D5 have way more heat pressure than any other pump at 3.9m and 1500L/h and DDC is only slightly more at 4.2m but only rated to flow 1000L/h which is 66% of what D5 flows. To my way of thinking slightly higher flowrate with little lower lift balances out against slightly less flowrate and little more lift rating pretty much balances their performance to be about equal. While majority of CLCs do not publish pump specs, the few that do are like H50 (Asetek) is only 40L/h with no lift spec, Asetek 545LC is 61L/h and Cool IT Eco III 34.2L/h. Alphacool AIO pumps are rated 72L/h or 0.6m and 100L/h or 1m lift depending model. ID Cooling CLC pumps are all rated 96-106L/h but one which is rated 1.2-1.3m lift with their Dashflow being 450L/h or 3m lift.

xDiVolatilX system is cooling just fine, so obviously 20-30% pump speed is enough.​

The D5 from EK gives 3.9m while the DDC gives 5.2m of head pressure. A high restriction 360 radiator will have almost 1psi restriction at 1gpm, and a high restriction waterblock will have 1.5psi at 1gpm or more so if you have 2 of those 360s, another 120mm rad and 2 waterblocks and some tubing and 90 degree angles you will be well above 5PSI at 1gpm.

Take a look at https://www.xtremerigs.net/2015/12/23/ek-xtop-revo-d5-pump-top-review/3/ . This is an older test and older pump top, but if you look at the flow vs psi restrictions at the different levels you will see that the the D5 needs quite a bit of speed to get pump pressure. You could also use the calculator you linked to and you will see that it estimates flows below 1gpm even with 2x480 medium restriction radiators (HW-Labs GTX 480 are aprox medium restriction) and a high restriction waterblock.

https://www.techpowerup.com/review/corsair-hydro-x-series-xd5-pump-reservoir-combo/6.html this is another one with a medium restriction loop. Look at the drop off down to just 70% of pump speed and then imagine if this was a high restriction loop.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
The D5 from EK gives 3.9m while the DDC gives 5.2m of head pressure. A high restriction 360 radiator will have almost 1psi restriction at 1gpm, and a high restriction waterblock will have 1.5psi at 1gpm or more so if you have 2 of those 360s, another 120mm rad and 2 waterblocks and some tubing and 90 degree angles you will be well above 5PSI at 1gpm.

Take a look at https://www.xtremerigs.net/2015/12/23/ek-xtop-revo-d5-pump-top-review/3/ . This is an older test and older pump top, but if you look at the flow vs psi restrictions at the different levels you will see that the the D5 needs quite a bit of speed to get pump pressure. You could also use the calculator you linked to and you will see that it estimates flows below 1gpm even with 2x480 medium restriction radiators (HW-Labs GTX 480 are aprox medium restriction) and a high restriction waterblock.

https://www.techpowerup.com/review/corsair-hydro-x-series-xd5-pump-reservoir-combo/6.html this is another one with a medium restriction loop. Look at the drop off down to just 70% of pump speed and then imagine if this was a high restriction loop.
Strange how lift / head pressure rating of DDC varies from site to site ranging from 4.0 to 5.2 meters. Guess it depends on what top is being used. Either we are both right or both wrong depending on what data we look at. ;) I agree, 2x 480 rad restriction and high restriction waterblock can drop flow to about 1gpm, but 1gpm is 227.1L/h which is still many times what CLCs flow .. and look how many peeps are using them. (doing that hurt so much I had to take a stiff hit of Glenfiddish ;) )

It's hard for me to accept Extreme Rig statements like "the 25% power setting is virtually useless" about D5 pump when most if not all CLCs are less than that, yet make up a huge share of consumer computer market now. While I almost hate CLCs I have to admit their performance over many years of use have proven they do an adequate job of cooling. (hurt so much I had to take another stiff hit of Glenfiddish ;) )

Not sure what we are going on here about. My initial statement was less than 100L/h is not good / borderline too slow, there is a dramatic improvement in cooling up to about 200L/h where the flow rate curve is beginning to flatten out as flow rates increases on up to 400-500L/h. Your data says basically the same thing.

Here's another reverence from Koollance say basically same thing we keep saying:

Finding a Pump​

Koolance offers several pumps of various specifications. The more cooling components added to a cooling loop, the stronger the pump needed to counter flow restriction. For a typical computer cooling loop with a 3-fan radiator and a few water blocks, any pump offered by Koolance should provide enough flow.

Flow rate tends to be over-emphasized in PC cooling. For the majority of loops, effective flow rates higher than 1.5-2.0 LPM (0.4-0.5 GPM) won't contribute much, if anything, to thermal performance. A reliable pump is important, as is making sure it's strong enough to keep adequate flow through your selected components. But for users looking to improve thermal performance, increasing radiator size and airflow is almost always more effective.

Keep in mind that the maximum flow rate listed for pumps is at zero static head pressure, while the maximum static head is at zero flow rate. That means the actual flow rate in a cooling system will usually be quite a bit lower than the pump's maximum specification.
https://koolance.com/how-to-build-a-water-cooled-pc
 

evhvis

Limp Gawd
Joined
Feb 12, 2021
Messages
326
Strange how lift / head pressure rating of DDC varies from site to site ranging from 4.0 to 5.2 meters. Guess it depends on what top is being used. Either we are both right or both wrong depending on what data we look at. ;) I agree, 2x 480 rad restriction and high restriction waterblock can drop flow to about 1gpm, but 1gpm is 227.1L/h which is still many times what CLCs flow .. and look how many peeps are using them. (doing that hurt so much I had to take a stiff hit of Glenfiddish ;) )

It's hard for me to accept Extreme Rig statements like "the 25% power setting is virtually useless" about D5 pump when most if not all CLCs are less than that, yet make up a huge share of consumer computer market now. While I almost hate CLCs I have to admit their performance over many years of use have proven they do an adequate job of cooling. (hurt so much I had to take another stiff hit of Glenfiddish ;) )

Not sure what we are going on here about. My initial statement was less than 100L/h is not good / borderline too slow, there is a dramatic improvement in cooling up to about 200L/h where the flow rate curve is beginning to flatten out as flow rates increases on up to 400-500L/h. Your data says basically the same thing.

Here's another reverence from Koollance say basically same thing we keep saying:

https://koolance.com/how-to-build-a-water-cooled-pc
The thing is the HWlabs GTX 480 are medium restriction and should be somewhere around 0.4psi while a high restriction radiator like the HWlabs GTS are somewhere around 0.9 or more. Basically having 2 GTS will give similar pressure drop as having 4 gtx radiators and would generally overwhelm a d5 pump if used with high restriction CPU and GPU blocks. The thing we are disagreeing upon is what a D5 is capable of running and at what speeds. The 1gpm is where you generally start to see little benefit from going faster, but it is not like the cooling is horrible at 0.5gpm. You are just sacrificing quite a bit of extra cooling performance, but it will still significantly outperform air and CLC unless you have too little radiator capacity.

The thing is a CLC is a small radiator and a waterblock with limited tubing so they can get away with lower pump quality. It's not like people can compare it with a decent pump anyways and the cooling is good enough for most. If the manufacturer knows their pump is weak then they probably make the rest of the system very low restriction at the cost of max performance and it works OK as a package.

A d5 at 25% will be OK with a low restriction radiator and low restriction waterblock, but there would most likely be a gain of 2-4 degrees by running it at 100%. A lot of custom loops are more complex, e.g. 1-2 radiators, CPU block and GPU block where the the blocks are generally the most restrictive parts. In some scenarios 25% is fine to run at so I agree with that the statement of 25% being useless isn't universal.

I do prefer Lagavulin 16 myself, but Glennfiddich is good stuff as well.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
The thing is the HWlabs GTX 480 are medium restriction and should be somewhere around 0.4psi while a high restriction radiator like the HWlabs GTS are somewhere around 0.9 or more. Basically having 2 GTS will give similar pressure drop as having 4 gtx radiators and would generally overwhelm a d5 pump if used with high restriction CPU and GPU blocks. The thing we are disagreeing upon is what a D5 is capable of running and at what speeds. The 1gpm is where you generally start to see little benefit from going faster, but it is not like the cooling is horrible at 0.5gpm. You are just sacrificing quite a bit of extra cooling performance, but it will still significantly outperform air and CLC unless you have too little radiator capacity.

The thing is a CLC is a small radiator and a waterblock with limited tubing so they can get away with lower pump quality. It's not like people can compare it with a decent pump anyways and the cooling is good enough for most. If the manufacturer knows their pump is weak then they probably make the rest of the system very low restriction at the cost of max performance and it works OK as a package.

A d5 at 25% will be OK with a low restriction radiator and low restriction waterblock, but there would most likely be a gain of 2-4 degrees by running it at 100%. A lot of custom loops are more complex, e.g. 1-2 radiators, CPU block and GPU block where the the blocks are generally the most restrictive parts. In some scenarios 25% is fine to run at so I agree with that the statement of 25% being useless isn't universal.

I do prefer Lagavulin 16 myself, but Glennfiddich is good stuff as well.
I pretty much agree with what you are saying .. except your guess that D5 on 2x rads and blocks increasing from 25% to 100% is only 2-4 degrees. My guess is it would be at least twice that. Maybe 30-35% to 100% would be 2-4 degree improvement. All the data I've see showed most improvement from 15-25% with significant improvement on up to about 35%, with each 10% increase from there on showing only a little improvement. Obviously there are many variables involved besides radiator and water block resistance, including fittings, elbows, length of lines, etc.

Lagavulin 16 is very good! I got a bottle of 18 yr old for Christmas. It's not bad. Had Balvenie 17 last year. That was nice too. Here's a sip for you. :D

 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
The D5 from EK gives 3.9m while the DDC gives 5.2m of head pressure. A high restriction 360 radiator will have almost 1psi restriction at 1gpm, and a high restriction waterblock will have 1.5psi at 1gpm or more so if you have 2 of those 360s, another 120mm rad and 2 waterblocks and some tubing and 90 degree angles you will be well above 5PSI at 1gpm.

Take a look at https://www.xtremerigs.net/2015/12/23/ek-xtop-revo-d5-pump-top-review/3/ . This is an older test and older pump top, but if you look at the flow vs psi restrictions at the different levels you will see that the the D5 needs quite a bit of speed to get pump pressure. You could also use the calculator you linked to and you will see that it estimates flows below 1gpm even with 2x480 medium restriction radiators (HW-Labs GTX 480 are aprox medium restriction) and a high restriction waterblock.

https://www.techpowerup.com/review/corsair-hydro-x-series-xd5-pump-reservoir-combo/6.html this is another one with a medium restriction loop. Look at the drop off down to just 70% of pump speed and then imagine if this was a high restriction loop.
I converted and added metric water column to Extreme Rigs' D5 pump psi to gpm chart. Makes it easier for those of us who use metric system. ;)
1643120583406.png
Source:​
 

evhvis

Limp Gawd
Joined
Feb 12, 2021
Messages
326
I pretty much agree with what you are saying .. except your guess that D5 on 2x rads and blocks increasing from 25% to 100% is only 2-4 degrees. My guess is it would be at least twice that. Maybe 30-35% to 100% would be 2-4 degree improvement. All the data I've see showed most improvement from 15-25% with significant improvement on up to about 35%, with each 10% increase from there on showing only a little improvement. Obviously there are many variables involved besides radiator and water block resistance, including fittings, elbows, length of lines, etc.

Lagavulin 16 is very good! I got a bottle of 18 yr old for Christmas. It's not bad. Had Balvenie 17 last year. That was nice too. Here's a sip for you. :D

The 25% was for a low restriction loop, not 2x medium restriction loops ;) Anyways thought you would like some hard numbers to go with the speculation :)

Finally got my Aquacomputer highflow next units and installed one in my CPU loop on my main. It is 2 fat 480 radiators which should be around medium restriction with a d5 pump and medium/high restriction water block, aprox 1m of tubing and only 1 90 degree angle. Did a quick check of temps vs flow and pump speed with fans at a fixed moderate speed (using maybe around 40% of the radiator+fans capacity).

Done on a 5900x with all core load of aprox 143w for more than 2 minutes and ambient between 22.9 and 23.0 for all runs at steady state. The uncertainty should be less than +/- 0.3 degrees as there was no changes other then pumpspeed and the average over 2 minutes has always been very consistent even though I run normal PB.

Pump at 100%
  • Die average: 56.1
  • Watertemp out of CPU block: 27.1
  • Flowrate: 266.8 l/h
Pump at 50%
  • Die average: 56.7
  • Watertemp out of CPU block: 27.3
  • Flowrate: 102.6 l/h
Pump at 27%
  • Die average: 58.7
  • Watertemp out of CPU block: 29.3
  • Flowrate: 31.6 l/h

Interestingly enough the water temp goes significantly up as the flow rate goes down which would make the radiator more effective due to increased differential between ambient and water. The block becomes much less effective though due to less turbulence. I ran at 27% due to 25% being below what the flow meter will read. It should also be noted that my waterblock is among the more insensitive to flow rate.

For comparison I do believe the h150i pro xt running with everything at max was somewhere between 64 and 67 degrees in the same heat test, but was sometime late in 2020 so don't remember the exact numbers. Mostly use the AIO to check that everything is working after building in a case as it is a pain to remove a custom loop, even with soft tubing.
 
Last edited:

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
Your results are basically what I've seen up to your 266.8L/h maximum flow rate. As your system tops out at 266.8L/h we can't see what differences would be above 266.8L/h.
How old is your pump?

I only used D5 for a few years. Before that it was whatever we could find/machine ourselves because there were no PC water cooling components on the market. When Thermalright stared making heat pipe tower coolers in about 2005-6 I began using more and more air coolers and (mostly tower coolers) and slowly quit using water.
 

evhvis

Limp Gawd
Joined
Feb 12, 2021
Messages
326
Your results are basically what I've seen up to your 266.8L/h maximum flow rate. As your system tops out at 266.8L/h we can't see what differences would be above 266.8L/h.
How old is your pump?

I only used D5 for a few years. Before that it was whatever we could find/machine ourselves because there were no PC water cooling components on the market. When Thermalright stared making heat pipe tower coolers in about 2005-6 I began using more and more air coolers and (mostly tower coolers) and slowly quit using water.
Pump is about 2 weeks old and 1gpm resistance of block + radiators should be in the 1.8 to 2.2 psi range with tubing and other stuff in addition. Difference in temp with regards to flow is very water block dependent. There probably aren't much benefits from going higher on flow rate in my system considering there is only around .6 degree difference between aprox 1.1gpm and 0.5gpm. The pump+res that was in my previous cooling setup is much older, but used the new pump+res right away as the old one will be used in another part of the cooling setup for my main rig and was due for cleaning.

The setup in my secondary system should have somewhere between 1 and 1.15 PSI resistance in block plus radiator at 1gpm which should allow significantly higher flow rates, but not going to put a flow meter in that as my second flow meter is for the main system as well and the highflow next meters are fairly expensive.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,479
Thanks for the info. While I do quite a bit of testing myself, much of what I know is results of others' work and reporting. Not enough time and/or money to get all the different components out there and test them all, so I stick to coolers (and some AIOs/CLCs), fans and cases now.
 

evhvis

Limp Gawd
Joined
Feb 12, 2021
Messages
326
Keep in mind that I am using a Techn AM4 block which is supposed to be among the least flow sensitive water blocks. I wouldn't be surprised if the gaps would be 1.5 to 2 as large in temps between the different flow rates if I was still using an EK Velocity block but max flow rate would be a bit higher.
 
Top