Overclock.net › Forums › Cooling › Water Cooling › Radiator size vs. Power vs. Temps. Please post your numbers!
New Posts  All Forums:Forum Nav:

Radiator size vs. Power vs. Temps. Please post your numbers!

post #1 of 8
Thread Starter 
I thought it would be neat to see how TDP and radiator count affects cooling taking TDP into consideration. This is a work-in-progress, so feedback is welcome on how to improve consistency.

If I get enough data, I'm planning to get some kind of chart built to look at TDP/Radiator vs. temperature. I suppose we could then think about questions like "What is the optimal radiator count for 300W TDP?
I've thought about this quite a bit, but I thought it'd be cool to open it up to all of you to see if we all huddle in a similar TDP/Rad ratio, and to see if going extreme actually helps. Hopefully you guys find this interesting.


Standardization:
For the sake of comparison, lets say a 120mmx120mm is 1 radiator unit. So if you have a 240, it'd be a 2, and a 360, it'd be a 3. If you have a Quad, and a Double, add them up to get a value of "6" RUs

Radiator considerations:
* 140mm square radiators have about 36% more surface area, but finding fans with sufficient Static Pressure (SP) is a challenge. Lets treat 140MM as 120, but please make a note on your post if you use 140mm.
* From my research, radiator thickness has little effect on cooling, so I'm going to ignore that. Don't get me wrong, there certainly is an effect, but its not big enough.

Temps:
I'd like to see one number with all your components stressed with AIDA64 Stress Test with the following settings:
* Stress FPU (Make sure CPU is UNchecked for peak thermal load)
* Stress GPU(s)
Run the test for 15 minutes and report max temperatures.

TDP:
I'd like to see standard TDP values, but please make a note if you overclock, and give me adjusted values if possible:

Video card TDP list: http://www.geeks3d.com/20090618/graphics-cards-thermal-design-power-tdp-database/
CPU TDP List: https://en.wikipedia.org/wiki/List_of_CPU_power_dissipation_figures

Radiator Fan configuration:

Bottom line:
At the minimum, I'd like to see 3 numbers:
* Total TDP of your water cooled components.
* Radiator Count
* Temps - both gaming and max (If you can only provide one, provide the max temp.)

*********************************
My computer's numbers:
I'll start this off with my components:
5930k CPU - 140W
980 Ti GPU - 250W
980 Ti GPU - 250W
Total TDP: 640W

Radiator: A single Triple 260 Rad, 3 units.

Total TDP/RU: 213W /RU
(This is how much power is being dissipated per unit area of radiator.)

Max Temps with AIDA64 Stress Test:
GPUs: 50C (reported by GPU-Z)
CPU: 62C (reported by AIDA64) Real Temp reported max temp of 65C on Core 2.

Fan Configuration:
3X Noctua NFF12 PPC
Sadly, my fans run on a temperature-sensitive curve, which is going to throw off the nubmers.
One fan is always on, but the other two turn off at temps below 40C. Even at 60C, they're not at 100%.

Moderators: I'm relatively new, so please feel free to edit/clean up my thread with any cool linking tricks etc.
Edited by Ziplocbox - 8/31/16 at 5:10pm
post #2 of 8
Way too many variables here - case, fans, setup, etc. for the numbers to be meaningful in the way intended. You also need specific parameters as a control for any testing. "While gaming" is totally random and as meaningful as "I like Pop Tarts" with no control - you need a fixed repeatable benchmark specified. Same goes for "Stress Test Load". Is that OCCT? AIDA64? IBT? P95? What settings?

It's a cool idea, but you have left way too much open ended. Tighten it up a bit and you can get some meaningful results.
Quote:
Originally Posted by Ziplocbox View Post

* From my research, radiator thickness has little effect on cooling, so I'm going to ignore that. Don't get me wrong, there certainly is an effect, but its not big enough.

I would say at 20% difference at moderate fan speeds with the difference increasing as fan speeds increase is a big enough effect......

post #3 of 8
Thread Starter 
Quote:
Originally Posted by ciarlatano View Post

Way too many variables here - case, fans, setup, etc. for the numbers to be meaningful in the way intended. You also need specific parameters as a control for any testing. "While gaming" is totally random and as meaningful as "I like Pop Tarts" with no control - you need a fixed repeatable benchmark specified. Same goes for "Stress Test Load". Is that OCCT? AIDA64? IBT? P95? What settings?

It's a cool idea, but you have left way too much open ended. Tighten it up a bit and you can get some meaningful results.
I would say at 20% difference at moderate fan speeds with the difference increasing as fan speeds increase is a big enough effect......

Agreed, I actually meant to take out the gaming benchmark part since that's pretty much nonsense. Edited.

I completely forgot about CPU stress test lol. Hmm, the best program I can think of to stress test BOTH CPU and GPU is AIDA64. I'll update the thread. The bottom line is to get 100% utilization on both CPU and GPU - which is tricky since 100% CPU usage can reduce GPU usage, so it needs to be standardized.

Regarding radiators, a few simplified tests that only vary the thickness seemed to hint at low variance:


Thinner radiators require reduced SP to push the air through, so I'm wondering if that has something to do with. Clearly fan performance is also wildly different, so its possible for a thin twin radiator design with two delta fans to do better than a triple rad with crappy fans etc. I'm hoping that a decent sample size will help with these variances. I'm looking for a line of best fit with many data points, so individual accuracy will iron itself out.

Thanks for the great feedback!
post #4 of 8
Too many unknowns,

My config:
6800k @ 4.5GHz, 160W
GTX 1080 OC 250W
Total 410W

Rads: 5, 120+240+280
Fans: 3 Vardar F1-120 @1150, 2 Vardar F1-140 @1150

Radio = 410/5 = 82WPR (watts per rad)

It only make sense to measure the water temp. To many variables for CPU and GPU temp readings.

At full load, handbrake+gaming over 1 hour

My water temp is at 53C and ambient temp is 24C

My three radiators a tightly fit in a small ITX Manta case. Temps will be better for sure if they are in a open air case.
post #5 of 8
Agreed...Lots of variables involved when water cooling to achieve your desired results.
Remember bigger rads over multiple small ones any day and to achieve really low deltas at low fan speeds more rad space the merrier

To give you a idea of the correct amount of heat to disipate at different fan speeds.Note it does not consider rad thickness and fpi as this only on average improves the cooling ability by 10-50 watts depending on your quailty of rad

Im at sub 10 delta at 1000-1200 rpm
i have a 480 monsta,360 ek and a 240 XSPC(all thick copper rads)
post #6 of 8
Not to mention the fact that there can be a great deal of disparity in rads of the same size from one to the next. For example, the EK 360SE can only dissipate ~75% as much heat as a Nemesis 360 GTS. That alone makes the entire "Radiator size vs. Power vs. Temps" a completely useless comparison when using those factors only.
post #7 of 8
Thread Starter 
Of course there will be a lot of difference between individual designs. For every guy with a thin radiator, there's a guy with a fat one. In the end, it will even out, but if and only if you have a large enough sample size.

There are an incredible number of variables that probably cause a lot more problems like:
When was the last time you cleaned your radiator?
What is ambient temperature?
Fan type (SP vs AF)
Blah blah...
I am looking for trends. It isn't possible to hold all variables constant... The question is whether you can get any meaningful data despite the limitations of the dataset.
post #8 of 8
From what I can understand, you want to see:

1. How much energy (power) a given radiator dissipate.
2. The temperature you get.

The temperature is affected by the water temperature, not directly by the radiator so if you want to test that, you must isolate that variable.
If the water temperature is constant, what you put in is dissipated out. The only change will be the water temperature that will rise with higher power input causing higher temperature.
If you don't monitor water temperature, this "survey" is worthless. I think that you would get more interesting data if you recorded the air temperature in and out with water temperature. With that you get pretty much everything you need to make more complex analysis.
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790k MSI Z97 Gaming 5 Sapphire R9 290X Tri-X 2x8gb ADATA 2133MHz CL10 
Hard DriveCoolingOSKeyboard
Mushkin 1 TB SSD & Samsung F3 spinpoint 1 TB MCP655A + 2 x MCP320 + EK supreme copper Win10 Corsair K70 Lux 
PowerCaseMouse
Corsair 750W Bitphenix Monstrous water cooling case Logitech G9X 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790k MSI Z97 Gaming 5 Sapphire R9 290X Tri-X 2x8gb ADATA 2133MHz CL10 
Hard DriveCoolingOSKeyboard
Mushkin 1 TB SSD & Samsung F3 spinpoint 1 TB MCP655A + 2 x MCP320 + EK supreme copper Win10 Corsair K70 Lux 
PowerCaseMouse
Corsair 750W Bitphenix Monstrous water cooling case Logitech G9X 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Water Cooling
Overclock.net › Forums › Cooling › Water Cooling › Radiator size vs. Power vs. Temps. Please post your numbers!