Recomended order for components (waterloop)

alc0

Honorable
Oct 5, 2012
54
0
10,630
Hi!

I am about to build my first waterloop, and I will cool both the CPU and the GPU in the same loop, with a D5 pump (MCP655) and 2x AX240 rads.

My thoughts are that the water from the GPU block will be hotter than the water from the rad, thus I should set it up as following:

Pump -> GPU-block -> Rad -> CPU-block -> Rad -> Res

My question is: will this give me any better temps than just have both the rads after each other? Like this:

Pump -> GPU-block -> CPU-block -> Rad -> Rad -> Res

It would be simpler to set up, but I will not sacrifice performance for simplicity.

Thanks in advance
AlC0
 
I do not do water cooling but common sense would indicate that you are 100% right about

gpu -> rad -> cpu -> rad - Res/pump.

Depending on what heats the water more, you may also be able to go

cpu -> rad -> gpu -> rad - res/pump

I await the water cooling experts for a better answer as well :)
 

ThomasJ93

Honorable
Apr 10, 2013
597
0
11,160
The order of your loop doesn't matter an awful lot. The difference in temperature only differs a bit usually. The reason for this is that the water flows so quickly, it barely has time to warm up. The only thing that is really important, is that your res is above the pump. That way it's nearly impossible for the pump to run dry.

Basically just do what you think looks best, as long as you keep the res/pump setup as i described it.
 

alc0

Honorable
Oct 5, 2012
54
0
10,630


That was the answer I was looking for. Thank you! Also, I am about to overclock my GPU and CPU as hard as I can, so every degree counts. Are you still sure that the order of the loop is (more or less) irrelevant?

I have calculated that my CPU (I5-3570K) will pull about 190 watts of power, wich gives out a maximum of 190 watts of heat ofcourse. And that the GPU (GTX560Ti) can pull a maximum of 250 watts. This means that I only have to dissipate 440 watts. Each of the rads should be able to dissipate about 400 watts (taken from the charts from pureoverclock.com), so with 2 of those I guess I can run on lower fan-speeds and get a great performance... what do you guys think?

AlC0
 

alc0

Honorable
Oct 5, 2012
54
0
10,630


Thats a good idea... I am pretty sure the GPU will be hotter, so I think i will go for that solution, if i don't end up with the rads after each other that is ;) thanks!
 

alc0

Honorable
Oct 5, 2012
54
0
10,630


What do you mean about the water equalizing? This is news to me, please elaborate :p

 


got to be a law of thermodynamics or something that states

you can only cool the water so far using the same components but in a different order

as the amount of water used, radiator surface , amount of fans used etc are all the same

therefor the temperature should equalise or stabilise the same

but as mentioned reservoir above the pump in either solution

 
If the waters heatup is that slow(due to how fast it moves) than yeah, the order will not matter at all since the water will pass each component many times before it even heats up by a degree.

So It looks like it will be more easy to setup after all.

Thinking of water like air must be a "my bad" idea.
 

alc0

Honorable
Oct 5, 2012
54
0
10,630



Interesting, I didn't know this, but it simplifies my building a great deal. Thanks man! And yes, have read about the res over pump many places, so thats a solid... won't fuck around pissing of gravity ;)
 

alc0

Honorable
Oct 5, 2012
54
0
10,630


Nice thread in your link. However, I have a couple of questions (hope you don't find this annoying):

1. The tests and setups is with only 1 heat source, is these results comparable with 2 heatsources?
2. 5-8 degrees seems significant to me? but maybe it isn't?

AlC0
 
the same bloke the link is quoting

martin--is top of his game when it comes to watercooling

and yes pretty sure hes done tests with more than one heat source

and i believe the 5c to 8c is referring to differences in parts of the loop regardless of which way round its done

not temps taken from the cpu or gpu

ie--leaving the radiator may be cooler than entering the radiator

again that makes sense to me--as i wouldnt expect totally even temperatures throughout the loop

as the radiators must dissipate more heat than the tubing
 

alc0

Honorable
Oct 5, 2012
54
0
10,630


Yes, this sounds precise and insightful. If I understand you correct, the 5-8 degrees will be before entering and after leaving both radiators after each other with both the heat sources in front of them like this:

CPU -> GPU -> delta T = 18 -> Rad -> Rad -> delta T = 10

then, if we should estimate how much impact this would have on each component it would be:

8/2 +- 25% (25% is the error margin to compensate for different heat output from each source.

8 degrees would be the conservative number.

Do I get this correct? Now it is more a question of understanding for me than actually what I need to know to build my loop, but I find this very interesting :)

AlC0