Originally Posted by akira749
I know that a loop reach equilibrium no matter where you put your rads versus your components.
Generally, I always put a rad before the pump so the water as just been "cooled". But in my next loop it might not be the case.
So my question is, will my pump suffer in any way if it receive "hot" water that has just pass thru 3 gpu, 1 cpu and 1 ram block without passing in any rads?
The water would then exit the pump, be cooled by a series of rads and then up to the components.
Equilibrium, in the manner **sometimes used** in water cooling, is a myth. Yes of course you will reach a point where the heat load in from the system and out thru the rads stops changing, but:
a) It takes quite a while, (contrary to a sometimes held popular belief, it's not a 10 minute thing) and
b) The temps do in fact change throughout the loop
Even under CPU benchmarks, I see 0.7C drop across the rads. With water sensors in and out of each rad, I see a minimum of 1.6C and as much as 3C depending on pump speeds running furmark. I don't think ya pump is going to be in any way affected by whether your water temp is 35 or 32C. OTOH, your CPU certainly will. Based upon testing, I have see as much as a 3C drop in CPU temps per 2C drop in water temps..... so if ya can get a Rad between ya GPUs and CPU could mean hitting ya self imposed temp limit or being just below it. Tho the reality there is....does the 2-3C really matter ?
Originally Posted by Vagrant Storm
But yeah, a water temp of 5-7° over ambient isn't horrible...not good and more radiator would help some, but it is a diminishing return. I run two 360mm radiators and my water temp is only about 4° over ambient. Heh...I can turn my fans off and it will only get to about 10° delta under normal loads.
I pulled this off martins some time back and it's what I generally go by (shooting for 10C) ..... of course using 10C as a target or base figure, makes all the math easy:
< 10C = Extreme WC System
10C High End WC System
20C Mid-range WC System
30C Entry level WC System
Of course normal loads will change per user so those numbers are intended for peak load, If you can maintain <10C under Furmark w/ twin GFX cards, you're doing perty well IMO.
Radiation / conduction is not an insignificant source of cooling....Look up what ya rads can do on Martins and then add up all ya heat sources. The result will be that most systems will handle maybe 50 to 75% of that load.....where is the rest going ? Energy can neither be created nor destroyed so it's gotta be going somewhere.
Under Furmark, my system is pulling 705 watts from the wall (635 watts of power / heat to the system) and the rads are handling 476 watts at 850 rpm according to martins tests .... radiation / conduction is handling the other 159 watts. Radiator shrouds are probably the biggest contributor as that heat loss is excluded from martins measurements, heat sinks next but every fitting, and other surface will radiate heat..... of course how much heat is pulled away is directly proportional to the movement of air across affected surfaces and the thermal conductivity of the material....GPU water blocks and backplates I expect are significant contributors also.
I find under CPU benchmarks, after 20 minutes, I am still changing..... under Furmark I get pretty close to stabilized after 20 minutes which is when I recorded the 1st set of numbers below. Just ran a test while typing this and chatting on phone w/ Son No. 1 down in Teneseeeeee at the moment:
Pump rpm =3200Fan rpm = 1185
Ambient = 23.4
420 In = 33.3
420 Out = 32.3
280 In = 32.7
280 out = 32.1
GPU 1 = 40.0
GPU2 = 40.0
Delta T Rads = 1.6C
Delta T Ambient = 8.7C
Pump rpm =3200Fan rpm = 850
Ambient = 22.9
420 In = 36.5
420 Out = 35.5
280 In = 35.9
280 out = 35.3
GPU 1 = 44.0
GPU2 = 44.0
Delta T Rads = 1.6C
Delta T Ambient = 11.6C
In 2nd test at lower fan rpm, was still climbing after 25 minutes (34.4C on 280 out....42C on GPU) and 0.1 higher just a minute later. At 30 minutes, made 34.7 and at 35 made 34.9 ... 38 min got us to 35.0....40 min got us 35.1 ..... at 42 min, still climbing and GPU temps just hit 44C ..... at 45 min still climbing (35.3C water temp) and I got bored ... ended the test and turned off the fans.
I think there is a value to a larger res in that the loop does take a long time to stabilize and that it is highly unlikely a typical user (non-synthetic scenario) will see a constant furmark like load over 20 or 45 minutes. The value of the larger thermal mass therefore is in equalizing or slowing temperature changes so that if you are using fan control (whether based upon water temps, GPU or CPU temps) you don't wind up chasing ya tail with the system responding to short term loadings.Edited by JackNaylorPE - 4/1/14 at 5:43pm