Custom Water Loops Temperature Problem

[Stocky]

Has anyone else realised that the custom water loops on the CPU are either worse or the same as the AIO?
The GPU's are off the rails good but right now there is no benefit what so ever to adding the CPU into the loop

[Joel Dash Reed]

Yep, I can't even OC a 7980XE to anything before hitting 100º C. Cooling GPUs and CPU. Or should only cool a single card? *sigh* I can't even pull more than 21k points on 3DMark.

[Turbo]

This is actually worse than I initially thought after reading this thread. After a few experiments, I've found that the numbers are WAY off.

Before I get into the numbers, I should say that all tests were done in a Define R6 and all used 4700 MHz RAM (though I don't think RAM matters here). AIO builds all used seven Cryorig QF140 fans (one in back, three up top, and two on bottom; yes the bottom front one clips the AIO, but that's besides the point) and a Corsair H150i Pro along with K|NGP|N 1080 Tis. All custom loop builds used the same CPU with a 360 rad up top and in front with a 140 rad on bottom along with the same watercooled 1080 Tis and acetyl/nickel CPU blocks. I also squeezed a Cryorig QF140 in the back exhaust slot.

Final note: all overclock values should match, system to system.

Now, without further ado, let's see just how wonky these numbers are.

Build 1: 1950X
OC Values -

• CPU: 4.05 GHz AIO; 4.05 GHz custom loop
  
• GPUs: 1925 MHz Core, 1975 MHz Memory air cooled; 1925 MHz Core, 1975 MHz Memory custom loop

AIO Temps -

• CPU: 66.04° C
  
• GPU1: 79.57° C
  
• GPU2: 83.24° C

Custom Loop Temps -

• CPU: 84.10° C
  
• GPU1: 89.61° C
  
• GPU2: 86.72° C

Build 2: 2700X
OC Values -

• CPU: 4.375 GHz AIO; 4.375 GHz custom loop
  
• GPUs: 2050 MHz Core, 2100 MHz Memory air cooled; 2050 MHz Core, 2100 MHz Memory custom loop

AIO Temps -

• CPU: 74.19° C
  
• GPU1: 87.07° C
  
• GPU2: 89.56° C

Custom Loop Temps -

• CPU: 91.31° C
  
• GPU1: 93.47° C
  
• GPU2: 90.11° C

Build 3: 7960X
OC Values -

• CPU: 4.3 GHz AIO; 4.3 GHz custom loop
  
• GPUs: 2010 MHz Core, 2085 MHz Memory air cooled; 2010 MHz Core, 2085 MHz Memory custom loop

AIO Temps -

• CPU: 67.51° C
  
• GPU1: 89.83° C
  
• GPU2: 92.39° C

Custom Loop Temps -

• CPU: 87.52° C
  
• GPU1: 94.63° C
  
• GPU2: 94.89° C

Build 4: 8086K
OC Values -

• CPU: 5.1 GHz AIO; 5.1 GHz custom loop
  
• GPUs: 2000 MHz Core, 2050 MHz Memory air cooled; 2000 MHz Core, 2050 MHz Memory custom loop

AIO Temps -

• CPU: 72.29° C
  
• GPU1: 83.38° C
  
• GPU2: 85.61° C

Custom Loop Temps -

• CPU: 84.55° C
  
• GPU1: 87.34° C
  
• GPU2: 88.14° C

As you can see, when comparing AIO/air temps to custom loop temps, the AIO/air temps trump the custom loop temps every time, especially in the case of the extreme chips. Further experimentation showed that custom loops that include only the CPU or only the GPUs resulted in vastly superior numbers along the lines of which I would expect a loop containing everything to meet, or even exceed.

Needless to say, something is very wrong here. I'm just speculating, but I'd assume the formula used to calculate the temperatures needs to be adjusted to better reflect a curve you'd see out of a custom loop system.

[Stocky]

The only time the custom loop comes in front is when you connect it in a multirad environment and leave the GPUs out of the loop. Because the temps are purely calculated with a simle formula with CFM of the CPU cooler and the frequency of the chip

They really need to fix this and go over the physics of overclocking as right now, spending 2 thousand dollars on a custom loop that yields worse results in 90% of situations is worthless

[Femboy Connoisseur]

Originally posted by CWY ^4Turbo:

This is actually worse than I initially thought after reading this thread. After a few experiments, I've found that the numbers are WAY off.

Before I get into the numbers, I should say that all tests were done in a Define R6 and all used 4700 MHz RAM (though I don't think RAM matters here). AIO builds all used seven Cryorig QF140 fans (one in back, three up top, and two on bottom; yes the bottom front one clips the AIO, but that's besides the point) and a Corsair H150i Pro along with K|NGP|N 1080 Tis. All custom loop builds used the same CPU with a 360 rad up top and in front with a 140 rad on bottom along with the same watercooled 1080 Tis and acetyl/nickel CPU blocks. I also squeezed a Cryorig QF140 in the back exhaust slot.

Final note: all overclock values should match, system to system.

Now, without further ado, let's see just how wonky these numbers are.

Build 1: 1950X
OC Values -

• CPU: 4.05 GHz AIO; 4.05 GHz custom loop
  
• GPUs: 1925 MHz Core, 1975 MHz Memory air cooled; 1925 MHz Core, 1975 MHz Memory custom loop

AIO Temps -

• CPU: 66.04° C
  
• GPU1: 79.57° C
  
• GPU2: 83.24° C

Custom Loop Temps -

• CPU: 84.10° C
  
• GPU1: 89.61° C
  
• GPU2: 86.72° C

Build 2: 2700X
OC Values -

• CPU: 4.375 GHz AIO; 4.375 GHz custom loop
  
• GPUs: 2050 MHz Core, 2100 MHz Memory air cooled; 2050 MHz Core, 2100 MHz Memory custom loop

AIO Temps -

• CPU: 74.19° C
  
• GPU1: 87.07° C
  
• GPU2: 89.56° C

Custom Loop Temps -

• CPU: 91.31° C
  
• GPU1: 93.47° C
  
• GPU2: 90.11° C

Build 3: 7960X
OC Values -

• CPU: 4.3 GHz AIO; 4.3 GHz custom loop
  
• GPUs: 2010 MHz Core, 2085 MHz Memory air cooled; 2010 MHz Core, 2085 MHz Memory custom loop

AIO Temps -

• CPU: 67.51° C
  
• GPU1: 89.83° C
  
• GPU2: 92.39° C

Custom Loop Temps -

• CPU: 87.52° C
  
• GPU1: 94.63° C
  
• GPU2: 94.89° C

Build 4: 8086K
OC Values -

• CPU: 5.1 GHz AIO; 5.1 GHz custom loop
  
• GPUs: 2000 MHz Core, 2050 MHz Memory air cooled; 2000 MHz Core, 2050 MHz Memory custom loop

AIO Temps -

• CPU: 72.29° C
  
• GPU1: 83.38° C
  
• GPU2: 85.61° C

Custom Loop Temps -

• CPU: 84.55° C
  
• GPU1: 87.34° C
  
• GPU2: 88.14° C

As you can see, when comparing AIO/air temps to custom loop temps, the AIO/air temps trump the custom loop temps every time, especially in the case of the extreme chips. Further experimentation showed that custom loops that include only the CPU or only the GPUs resulted in vastly superior numbers along the lines of which I would expect a loop containing everything to meet, or even exceed.

Needless to say, something is very wrong here. I'm just speculating, but I'd assume the formula used to calculate the temperatures needs to be adjusted to better reflect a curve you'd see out of a custom loop system.

Great work! Its a shame custom loop temps dont work the way they should, because the system for it so far is pretty fun.

How are you getting loops with just the GPUs? Whenever I try, I can't plug into the cpu cooler and the pump at the same time

[Turbo]

Originally posted by Natsu Dragneel:

How are you getting loops with just the GPUs? Whenever I try, I can't plug into the cpu cooler and the pump at the same time

You know what, now that you say that, I don't think I got it to work either. I was on some pretty heavy duty meds the night that I tried to do that, so I guess I'm not surprised I didn't remember that it didn't work. Though, that being said, I just realized that a fanless HE02 might do the trick. I'm going to go try it.

Edit:
It works! The CPU TPD is extremely limited, but I pulled it off. 8600K at stock clocks with two 1080 Tis overclocked to 2100 MHz Core, 2200 MHz Memory. Temps are: 77.83° C on the CPU, 88.04° C on GPU1 and 86.63° C on GPU2. At those temps, I could probably push the CPU up a couple of ticks if I wanted to.

It's in a Define R6 with a 360mm rad up top and 280mm rad on bottom with two Cryorig QF140s up front and a single in the back. Photos at the link.

GPU Only Custom Loop[imgur.com]

Just had another thought, would underclocking a CPU accomplish the same thing?

Edit 2:
NOPE! Looks like the temp curve doesn't extend to below stock clock speeds. Even lowering an 8086K to 3.0 GHz didn't do it.

[🦊Λℚ𝓤ΛƑΛᗯҜᔕ🦊]

Folks.. the developers already know nothing works right at the moment, it's early access. They'll re-visit all the cooling system later to actually make it cool correctly. They're just trying to actually add water cooling to the game at the moment. During this stage of EA they're just trying to get it in-game and working, no one's concerned about how well it cools or doesn't cool, it's EA. At least it's in the game and looks cool right now. The majority of the temperatures on most CPU's when overclocked don't work right in this game as-is. At least not for the high end parts.

I was able to play with a custom water loop and get a older quad core 6600K to clock higher on custom water than I've ever seen in PCBS on any air cooler or AIO, 4.95 Ghz in my testing. The most I'd ever had before for any 6600K (even after swapping out about 50 of em) was 4.7 ghz even with an AIO.

You're all just focusing on these higher end parts and thinking that's the whole picture and "OMG IT DOESN'T WORK BECAUSE IT WON'T COOL A 18 CORE CPU BETTER" etc. I guess most of y'all don't realize just how hot these things run. Go look at the Intel 18 core overclocking on youtube and see for yourself. Runs 100c in real life with just +200 mhz over even with a custom water loop with 360mm radiator. Also 8600K and 8700K are also stupidly hot chips even in IRL and will easily run 100c even under custom water loops just at 4.9 - 5.0 ghz. It's no surprise they run hot too in PCBS.

You're all just making a big fuss and ruckus over nothing. They'll revisit cooling later when the game's done, it's not even done yet and you're already getting your panties in a twist.

[Turbo]

Originally posted by Impending Rentacle Tape:

Folks.. the developers already know nothing works right at the moment, it's early access. They'll re-visit all the cooling system later to actually make it cool correctly. They're just trying to actually add water cooling to the game at the moment. During this stage of EA they're just trying to get it in-game and working, no one's concerned about how well it cools or doesn't cool, it's EA. At least it's in the game and looks cool right now. The majority of the temperatures on most CPU's when overclocked don't work right in this game as-is. At least not for the high end parts.

I was able to play with a custom water loop and get a older quad core 6600K to clock higher on custom water than I've ever seen in PCBS on any air cooler or AIO, 4.95 Ghz in my testing. The most I'd ever had before for any 6600K (even after swapping out about 50 of em) was 4.7 ghz even with an AIO.

You're all just focusing on these higher end parts and thinking that's the whole picture and "OMG IT DOESN'T WORK BECAUSE IT WON'T COOL A 18 CORE CPU BETTER" etc. I guess most of y'all don't realize just how hot these things run. Go look at the Intel 18 core overclocking on youtube and see for yourself. Runs 100c in real life with just +200 mhz over even with a custom water loop with 360mm radiator. Also 8600K and 8700K are also stupidly hot chips even in IRL and will easily run 100c even under custom water loops just at 4.9 - 5.0 ghz. It's no surprise they run hot too in PCBS.

You're all just making a big fuss and ruckus over nothing. They'll revisit cooling later when the game's done, it's not even done yet and you're already getting your panties in a twist.

This seems not helpful at all. Yes, watercooling works, just not well. I posted my reply to give the devs some real numbers and to show, not that it doesn't work, because it does, but that AIO and air cooling work better at the moment.

Also, yes, I understand it's early access, but the point of early access is to get feedback on what does and doesn't work in order to make the final product a more polished and refined experience. If we don't complain about the problems, how will they know to fix them?

[Stocky]

Originally posted by Impending Rentacle Tape:

Also 8600K and 8700K are also stupidly hot chips even in IRL and will easily run 100c even under custom water loops just at 4.9 - 5.0 ghz. It's no surprise they run hot too in PCBS.

Not true at all actually. I have built plenty of systems with 8600k and have no issue clocking them to 5.0-5.2ghz on air
They can overclock extremely well, in saying that I got an 8600k clocked to 5.7Ghz on a custom loop in PCBS with triple radiator setup but no GPU in the loop, the water cooling works okay as it is, they just need to dial down the heat added to the loop from the GPU's because its unrealistic by a long shot. They just need to implement a more complex formula for heat calculation and take into consideration a lot more other then CFM of the radiator and clock speed

[🦊Λℚ𝓤ΛƑΛᗯҜᔕ🦊]

Originally posted by Stocky:

Originally posted by Impending Rentacle Tape:

Also 8600K and 8700K are also stupidly hot chips even in IRL and will easily run 100c even under custom water loops just at 4.9 - 5.0 ghz. It's no surprise they run hot too in PCBS.

Not true at all actually. I have built plenty of systems with 8600k and have no issue clocking them to 5.0-5.2ghz on air
They can overclock extremely well, in saying that I got an 8600k clocked to 5.7Ghz on a custom loop in PCBS with triple radiator setup but no GPU in the loop, the water cooling works okay as it is, they just need to dial down the heat added to the loop from the GPU's because its unrealistic by a long shot. They just need to implement a more complex formula for heat calculation and take into consideration a lot more other then CFM of the radiator and clock speed

Okay, I was off a little bit. It's 5.0 ghz and above and all of them in to the 90's and above C.
5.2 ghz is the top 0.1% of 8700K chips and most won't do that. 5.0 - 5.1 is typical. See here:
https://www.youtube.com/watch?v=GRviKkVUAa0

[Jamala007]

Originally posted by Shoutout to Sean Ranklin:

Originally posted by CWY ^4Turbo:

This is actually worse than I initially thought after reading this thread. After a few experiments, I've found that the numbers are WAY off.

Before I get into the numbers, I should say that all tests were done in a Define R6 and all used 4700 MHz RAM (though I don't think RAM matters here). AIO builds all used seven Cryorig QF140 fans (one in back, three up top, and two on bottom; yes the bottom front one clips the AIO, but that's besides the point) and a Corsair H150i Pro along with K|NGP|N 1080 Tis. All custom loop builds used the same CPU with a 360 rad up top and in front with a 140 rad on bottom along with the same watercooled 1080 Tis and acetyl/nickel CPU blocks. I also squeezed a Cryorig QF140 in the back exhaust slot.

Final note: all overclock values should match, system to system.

Now, without further ado, let's see just how wonky these numbers are.

Build 1: 1950X
OC Values -

• CPU: 4.05 GHz AIO; 4.05 GHz custom loop
  
• GPUs: 1925 MHz Core, 1975 MHz Memory air cooled; 1925 MHz Core, 1975 MHz Memory custom loop

AIO Temps -

• CPU: 66.04° C
  
• GPU1: 79.57° C
  
• GPU2: 83.24° C

Custom Loop Temps -

• CPU: 84.10° C
  
• GPU1: 89.61° C
  
• GPU2: 86.72° C

Build 2: 2700X
OC Values -

• CPU: 4.375 GHz AIO; 4.375 GHz custom loop
  
• GPUs: 2050 MHz Core, 2100 MHz Memory air cooled; 2050 MHz Core, 2100 MHz Memory custom loop

AIO Temps -

• CPU: 74.19° C
  
• GPU1: 87.07° C
  
• GPU2: 89.56° C

Custom Loop Temps -

• CPU: 91.31° C
  
• GPU1: 93.47° C
  
• GPU2: 90.11° C

Build 3: 7960X
OC Values -

• CPU: 4.3 GHz AIO; 4.3 GHz custom loop
  
• GPUs: 2010 MHz Core, 2085 MHz Memory air cooled; 2010 MHz Core, 2085 MHz Memory custom loop

AIO Temps -

• CPU: 67.51° C
  
• GPU1: 89.83° C
  
• GPU2: 92.39° C

Custom Loop Temps -

• CPU: 87.52° C
  
• GPU1: 94.63° C
  
• GPU2: 94.89° C

Build 4: 8086K
OC Values -

• CPU: 5.1 GHz AIO; 5.1 GHz custom loop
  
• GPUs: 2000 MHz Core, 2050 MHz Memory air cooled; 2000 MHz Core, 2050 MHz Memory custom loop

AIO Temps -

• CPU: 72.29° C
  
• GPU1: 83.38° C
  
• GPU2: 85.61° C

Custom Loop Temps -

• CPU: 84.55° C
  
• GPU1: 87.34° C
  
• GPU2: 88.14° C

As you can see, when comparing AIO/air temps to custom loop temps, the AIO/air temps trump the custom loop temps every time, especially in the case of the extreme chips. Further experimentation showed that custom loops that include only the CPU or only the GPUs resulted in vastly superior numbers along the lines of which I would expect a loop containing everything to meet, or even exceed.

Needless to say, something is very wrong here. I'm just speculating, but I'd assume the formula used to calculate the temperatures needs to be adjusted to better reflect a curve you'd see out of a custom loop system.

Great work! Its a shame custom loop temps dont work the way they should, because the system for it so far is pretty fun.

How are you getting loops with just the GPUs? Whenever I try, I can't plug into the cpu cooler and the pump at the same time

I would first ask, did you do enough research to make sure the "law of diminishing returns" did not play a role in your 3-4 rad loop? Because it is a fact that overkill can indeed cause higher temperatures. Perhaps the power required to run the loop is causing increased temperatures into the loop. one thing you can try, invest in another pump, and separate the GPU's from the main loop thereby giving them their own loop exhausting to one of the other rad's. Or even separate the GPU's, perhaps one with the CPU and one alone.

[🦊Λℚ𝓤ΛƑΛᗯҜᔕ🦊]

Originally posted by Jamala007:

I would first ask, did you do enough research to make sure the "law of diminishing returns" did not play a role in your 3-4 rad loop? Because it is a fact that overkill can indeed cause higher temperatures. Perhaps the power required to run the loop is causing increased temperatures into the loop. one thing you can try, invest in another pump, and separate the GPU's from the main loop thereby giving them their own loop exhausting to one of the other rad's. Or even separate the GPU's, perhaps one with the CPU and one alone.

Check the date on the replies before you reply to a thread next time. This is a 5 month old thread from back when the game was in early access even before final release. Nothing in this thread is even remotely relevant to anything to do with PCBS today. Where did you even manage to find this?

Also it's pretty obvious by your post you have no idea how cooling works in PCBS. "Law of diminishing returns" may exist in real life, but it does not exist in PCBS. More radiators = more cooling with no limit in PCBS.

[Squall Leonhart]

Originally posted by Aquafawks:

Originally posted by Jamala007:

I would first ask, did you do enough research to make sure the "law of diminishing returns" did not play a role in your 3-4 rad loop? Because it is a fact that overkill can indeed cause higher temperatures. Perhaps the power required to run the loop is causing increased temperatures into the loop. one thing you can try, invest in another pump, and separate the GPU's from the main loop thereby giving them their own loop exhausting to one of the other rad's. Or even separate the GPU's, perhaps one with the CPU and one alone.

Check the date on the replies before you reply to a thread next time. This is a 5 month old thread from back when the game was in early access even before final release. Nothing in this thread is even remotely relevant to anything to do with PCBS today. Where did you even manage to find this?

Also it's pretty obvious by your post you have no idea how cooling works in PCBS. "Law of diminishing returns" may exist in real life, but it does not exist in PCBS. More radiators = more cooling with no limit in PCBS.

who are you kidding, the custom loop still doesn't work properly.

[🦊Λℚ𝓤ΛƑΛᗯҜᔕ🦊]

Originally posted by Squall Leonhart:

who are you kidding, the custom loop still doesn't work properly.

Custom loops in PCBS do not work like they do in real life, this is true. You have to "understand" how PCBS does water cooling. In PCBS like I said above, more radiators = more cooling. You get better cooling in PCBS for example with 7 x 240mm radiators compared to 5x360mm radiators. Even though in real life the 5x360mm option would cool better.