Advice on loop and pump feasibility

[Jim Morbid]

Hi experts

I'm putting together a loop, which originally was meant to be a dual loop (GFX and CPU/Chipset) but I'm rapidly running out placement space and am thinking about consolidating the two loops into one.

As it stands, I'm thinking of this: Petra'ed Laing DCC -> MCR320 -> MCR220 -> Apogee GTZ -> Chipset (dont know yet) -> MCW60 -> MCW60 -> Res

I've seen a few people run this kind of loop on a single pump but I would like some input from the vets.

Thanks,

JM

[Kibbler]

I would do...

Petra'ed Laing DCC -> MCR320 -> Apogee GTZ -> Chipset (dont know yet) -> MCR220 -> MCW60 -> MCW60 -> Res

You lose efficiency by putting the rads together. In your example, the MCR220 would be trying to cool down, you know, already cool water. Split them up so that each rad has a water/ambient temp delta to work with.

[Giannis86]

im afraid you are wrong.loop order barely makes any difference overall in the loop. The water tries to reach thermal equilibrium so the placement of the rad wont matter. The only thing it will affect is the pressure drop it has.

[Jim Morbid]

im afraid you are wrong.loop order barely makes any difference overall in the loop. The water tries to reach thermal equilibrium so the placement of the rad wont matter. The only thing it will affect is the pressure drop it has.

Thats what I thought. Could you elaborate on the "The only thing it will affect is the pressure drop it has" part please?

Thanks,

JM

[Kibbler]

Hmm. My mistake. Learn something new every day.

[septim]

water pressure benefits some blocks... restrictive blocks mostly...

So for the first few parts of your loop: res-pump-rad-CPU block-other parts...
We see the stronger pressure from pump to rad to cpu block.
Some cpu block benefits from stronger pressure more than others though. EK supreme is a good example.

[Jim Morbid]

Ahh no I understand the statement. Placement only matters when your trying to maximize the pressure over a certain component? IE, if I was using a Supreme it would make sense to up it first in the loop so that it has 'maximum' pressure.

Thanks for clearing that up.

[Giannis86]

i would actually recommend the same.
i have 3 of these babies (2 installed at the moment). the perform excellent and they bleed very fast.
last week when i installed a 2nd loop for my 4870x2 it was fully bled within 1-2 hours..

as about that pressure drop thing you got it right.i am not quite sure how the gtz compares with the ek supreme in this situation. personally i have my fuzion v2 right after my pump as the 360gtx rad is more restrictive than other rads and i felt that the fuzion would benefit being before the rad (havent tested the other way though so im not sure.

the gtz is slightly more restrictive so i would make sure it before all other blocks in the loop (unless it creates a big mess with the tubing..)

[MpG]

I'm not sure any modern waterblock is more restrictive than the Supreme. Even it's biggest fans admit that much. :D The GTZ's pressure drop is a little higher than the Fuzion V2, but still pretty reasonable.

I've never seen evidence that the loop order alters the pressure drop across a component. If it did, then you'd want the critical component placed immediately after the pump, but I have yet to see any demonstrated evidence of such.

[StAndrew]

im afraid you are wrong.loop order barely makes any difference overall in the loop. The water tries to reach thermal equilibrium so the placement of the rad wont matter. The only thing it will affect is the pressure drop it has.

I heard this stated more times than one and here are my thoughts (please correct me if im wrong):

(This is all an assumption based off my limited knowledge of thermodynamics and water loops :rolleyes:). If you keep your system at one single load, emitting a specific amount of heat per unit time, then yes, you would reach an equilibrium. However, thats a far fetch from how most systems work. As your CPU and GPU change from a idle state (specially when most CPUs and GPUs go into a "deep sleep" when idle for power saving reasons) to a load state (specially a heavy load), I figured it would be marginally beneficial to have your rad cooling spread out (specially if your CPU and GPU are reaching load temps independent of each other, ie high CPU load with low GPU load and vice versa). I would assume that since most of today's computer parts are designed to go into many different levels of "sleep" in order to save power, it would be rather difficult for one to consistently find an equilibrium in ones system usage (except during game play or other predictable and long running programs). Of course if you are right, it wouldn’t matter where the rad is placed anyways :).

Of course, with all this said, I think it is also important that you focused your loop design in a more direct manner, eliminated un-needed loops/bends, keep the tube length as short as possible (to a point, don’t make it too short as to cause kinking at fittings), and creating the most straight forward flow path possible.

Good luck Morbid :up:

[Jim Morbid]

I heard this stated more times than one and here are my thoughts (please correct me if im wrong):

(This is all an assumption based off my limited knowledge of thermodynamics and water loops :rolleyes:). If you keep your system at one single load, emitting a specific amount of heat per unit time, then yes, you would reach an equilibrium. However, thats a far fetch from how most systems work. As your CPU and GPU change from a idle state (specially when most CPUs and GPUs go into a "deep sleep" when idle for power saving reasons) to a load state (specially a heavy load), I figured it would be marginally beneficial to have your rad cooling spread out (specially if your CPU and GPU are reaching load temps independent of each other, ie high CPU load with low GPU load and vice versa). I would assume that since most of today's computer parts are designed to go into many different levels of "sleep" in order to save power, it would be rather difficult for one to consistently find an equilibrium in ones system usage (except during game play or other predictable and long running programs). Of course if you are right, it wouldn’t matter where the rad is placed anyways :).

Of course, with all this said, I think it is also important that you focused your loop design in a more direct manner, eliminated un-needed loops/bends, keep the tube length as short as possible (to a point, don’t make it too short as to cause kinking at fittings), and creating the most straight forward flow path possible.

Good luck Morbid :up:

I agree with you, but I think the gains are so small that it's not worth the time (and Tygon) to make it work. Plus, you're only truly worrying about temps when your system is working at full steam. Not going to loose sleep over an idle temp that could be 4degree's better :P

Thanks for your input.

JM

[MpG]

If you keep your system at one single load, emitting a specific amount of heat per unit time, then yes, you would reach an equilibrium. However, thats a far fetch from how most systems work. As your CPU and GPU change from a idle state (specially when most CPUs and GPUs go into a "deep sleep" when idle for power saving reasons) to a load state (specially a heavy load), I figured it would be marginally beneficial to have your rad cooling spread out (specially if your CPU and GPU are reaching load temps independent of each other, ie high CPU load with low GPU load and vice versa). I would assume that since most of today's computer parts are designed to go into many different levels of "sleep" in order to save power, it would be rather difficult for one to consistently find an equilibrium in ones system usage (except during game play or other predictable and long running programs). Of course if you are right, it wouldn’t matter where the rad is placed anyways :).
A well designed WC system is tested by putting the computer at maximum load, which IS is predictable program. So in that case, equilibrium is definitely reached. While it's not indicative of normal usage, it can be safely said that under normal usage, temps will never reach the equilibrium that was reached under maximum. Therefore, planning a system to minimize what temperature that 'equilibirum' occurs at, isn't a bad idea at all.

[Giannis86]

you do have a point there.however PERSONALLYi care more about load temps of the system as a whole (lets say a game for example). My idle temps are low enough that a few degrees wont bother me (passive rad even worked for me for idle).

[StAndrew]

A well designed WC system is tested by putting the computer at maximum load, which IS is predictable program. So in that case, equilibrium is definitely reached. While it's not indicative of normal usage, it can be safely said that under normal usage, temps will never reach the equilibrium that was reached under maximum. Therefore, planning a system to minimize what temperature that 'equilibirum' occurs at, isn't a bad idea at all.

What about this (all hypothetical):

Your CPU puts out lets say x btu's at load. You have a 120-240mm rad that removes 2xbtu's. If you are running two GPU's after your CPU, wouldnt it be very beneficial to put the 120mm rad between the CPU and GPUs? You would be effectively eliminating any heat from the CPU before it reaches the GPUs. If you have a 240-360 rad after the GPUs, able to eliminate all heat created by the GPU's, wouldnt it be safe to assume that your CPU and GPU would operate as if they were on seperate loops ('cept for the pump).

[MpG]

That's correct. However, the temperature rise as the water goes through a waterblock is almost always under 1 degree, unless you've got an insanely hot component (you don't), or dismally low flow (you shouldn't), so even if you don't stagger the rads, the worst case scenario would see the later components receiving water 1 degree hotter than normal. Since the one degree is highly unlikely to make much of a difference to the GPU, it kind of becomes a moot point. Much better to arrange the components in the cleanest, most efficient way possible.

[majestik]

Why not go with a "better" top for your DDC?? Something that performs better, like the XSPC Reservoir Top...:D
Because Delrin > Acrylic.

I c wut u did thar with quotation marks on better, though :D

[majestik]

What about this (all hypothetical):

Your CPU puts out lets say x btu's at load. You have a 120-240mm rad that removes 2xbtu's. If you are running two GPU's after your CPU, wouldnt it be very beneficial to put the 120mm rad between the CPU and GPUs? You would be effectively eliminating any heat from the CPU before it reaches the GPUs. If you have a 240-360 rad after the GPUs, able to eliminate all heat created by the GPU's, wouldnt it be safe to assume that your CPU and GPU would operate as if they were on seperate loops ('cept for the pump).

What people don't seem to understand is that difference between "Water in" and "Water out" for a radiator is less than 2C.

See Water In and Water Out sensors
http://www.martinsliquidlab.com/img/TFC480-Thermal1.png

(courtesy of martin)

[IanY]

I can't tell you what to do, and what you end up doing should entirely be your choice.

However, I will tell you what I would do. I would split the cpu from the rest. Its the cpu that you would want to cool best, and the actual tempsof the NB and GPUs are somewhat less important than the cpu.

I would take the MCR220 or the MCR320, take the Apogee GTZ, and then buy another pump and isolate that loop from all the other "stuff" (trying to be polite). In that way, the heat from the NB and GPUs willnot influence the load temps of the cpu.

Again, its all up to you.

[StAndrew]

That's correct. However, the temperature rise as the water goes through a waterblock is almost always under 1 degree, unless you've got an insanely hot component (you don't), or dismally low flow (you shouldn't), so even if you don't stagger the rads, the worst case scenario would see the later components receiving water 1 degree hotter than normal. Since the one degree is highly unlikely to make much of a difference to the GPU, it kind of becomes a moot point. Much better to arrange the components in the cleanest, most efficient way possible.

This is true. I think the most important aspect of setting up a loop is making it as efficient as possible. However, I like to have the Rads staggered so I can adjust the fans on my rads for accommodating heavy CPU loads or GPU loads (ends up actually working the best for my loop anyways in terms of effectiveness).

Thanks for that graph Majestik. I was just a bit confused as I keep hearing rad placement makes NO difference, however common sense would suggest that there should be at least a marginal difference in your setup. However, I agree its not that much of a diff to try and force a staggered setup and subsequently reducing your loop efficiency.