If I'm counting right...you have 8+ waterblocks and two radiators? You need more than one pump.

one loop is easier to maintain, and most will argue its just as effective. but really a matter of personal preference.

but you need more than one pump no matter what you do. i would run them in series. i'll bet your flow is less than 0.3 gpm with only one pump and all those blocks.

I'm building in a TH10 also right now. You might get some ideas from my build log. http://www.overclock.net/t/1266202/build…a-the-bbbb/0_20

As far as guesstimating the flow...I can help you take a better shot at it. Electrocutor (member here) helped me undertstand the pressure curves better....

Your D5 has a pressure curve like this:




Depending on which pump top you have, ....if you were targeting a flow rate of 1.0 gpm (which is the minimum you should target), you would need head pressure of less than 5 PSI in order to hit that with one pump. So...compare the 5 PSI target to the resistance in the blocks you are using, and you'll see if you come close. Not all manufacturers publish the info, so you'll have to look to reviews to get it. I'll start you off with a format to use...and you can search and fill in the actual #'s:

Kryos Waterblock @ 1.0 gpm = 1.5 psi
Kryos Waterblock @ 1.0 gpm = 1.5 psi
AquagraFX block @ 1.0 gpm = 2.0 psi
MIPS chipset block @ 1.0 gpm = 1.0 psi
MIPS Raid card block @ 1.0 gpm = 1.0 psi
MIPS SAS expander block @ 1.0 gpm = 1.0 psi
AQ5 block @ 1.0 gpm = 1.0 psi
AQ5 block @ 1.0 gpm = 1.0 psi
RAM block @ 1.0 gpm = 1.0 psi
RAM block @ 1.0 gpm = 1.0 psi
Radiator 1 @ 1.0 gpm = 1.0 psi
Radiator 2 @ 1.0 gpm = 1.0 psi

Total psi of 14. One pump simply wouldn't keep you at 1 gpm...under even the best of scenarios. You should search for those blocks where you can get some actual information...or at least estimate better based on other brands. Also...there are more restrictive and less restrictive blocks/rads out there...so you can either adjust the blocks/rads you are using to be the least restrictive possible....or you can go get another pump or two.

I would go more pumps, and put them in series. You can buy bigger more powerful pumps....but I love the D5 because it is quiet and looks really nice when you get a dress-up mod kit for it.

I'm using 4 D5's in my build.

Oh boy, here we go again...........

First, I am experianced in fluid flow and thermodynamics.

Second, Why are you water cooling your system? If it is to reduce noise, get enough radiator capacity to remove 100 to 125% of the capacity of your power supply. The more radiator surface, the more fans you can use, and the slower, and quieter, the fans will run. If you are intending to overclock the CPUs, then you have to decide if you want the lowest possiable temperatures of the CPUs, or just low enough to be stable in operation.

For any situation, 1 gpm is way high for flow. A Good, comfortable and practical flow is about 1/3 of that, or 1 L/hr. The lower flow will still remove all the heat you can possiably generate in your CPU and other components, and allows less pump power. The higher flow does allow slightly lower temperatures, on the order of 1 degree Centegrade, but does not remove more heat. Remember, heat is not equal to temperature.

My sytem operates at 0.6 L/m, and maximum CPU and GPU die temps are between 60 and 65 C, and chip service life starts to degrade at temps above 95C. Plenty of room for error. Increasing flow to 1.5 L/min does not reduce temps on the chips by 50%, it drops them by about 3C, and water temp by about 1C.

A small trickle of water flow will be more than enough to cool everything like Motherboard,RAM and probably your raid controllers. (I am not familiar with what you are talking about with them) Basically, I suggest that if you connect everything except CPU and GPU in one loop with the Aquastream pump you have, and then get a Aquacomputer D5 for the other stuff, you will be just fine. Use a small radiator for the RAM etc, and all of the rest for the CPU/GPU loop.

Use a small tube for the RAM/MB loop. The Aquacomputer 6mm ID tube and fittings are very easy to work with and will give plenty of flow with the Aquastream pump. Use bigger tube for the CPUs and GPU blocks if you want. I use an 8mm ID tube with an AC D5 pump and get a max flow over 2 L/min. Which is just fine.

Now that I have made my suggestions, I will go lurk again. I have already said enough to start a flame war, and I don't want to do that again.

One other thing to consider, if what you start with is too small to keep temps under control, you can always add pumps or radiators or fans later. Starting with more than you need is overkill and wasted money.

One other point. I was taught to call them 'Pump Laws'

If you double the pressure drop in a closed loop system, you cut the flow rate in half.

In order to double the flow rate, ie 1 gpm to 2 gpm, it takes approximatly four times the pump head, and eight times the pump power.

Don't worry. No flame war starting. But telling him he will be fine at 0.3 gpm is not conventional wisdom.

I would suggest gobase2008 go check out results by independent reviewers, and get their take on optimal flow rate. Martin's Liquid Labs suggests 1.0 gpm being the right target, and Skinnee Labs suggests that 1.5 gpm is optimal. They test scores of blocks and pumps, and their suggestions are not based on their experience with one or two machines.

And those "pump laws" may work out close to reality in some situations...but of course the right answer is to look at the pressure curve for the pump and pump top that has been selected. Different pumps and different tops have different pressure curves.

gobase...here is some actual test data comparing different flow rates and their impact on temperature. These types of curves are fairly consistent anywhere I've seen independent testing done. It will depict why I think you need a much higher target than 0.3 gpm. Some blocks create as much as an 8 degree temp difference just be increasing flow rate. And that's a big deal. Good luck with your build!

I'll add my bit and then be quiet ...

They are called 'laws' because they do not vary anywhere in the universe (except when conditions approach the speed of light) and have been derived/discovered through science and repeated experiment to test the hypotheses behind them. The devices used in PC watercooling obey the same laws however I would say there is a much higher degree of error in the software and monitoring equipment than in larger and more expensive industrial setups with mission critical precision and accuracy. My Intel 980X, for example, reports core temperatures at idle that are 6 to 10 degrees below ambient temperature in five out of six cores, which does not make sense, so I cannot ever use that data as trustworthy while monitoring my cooling loop.

Although I am not a fluid dynamics expert I have a science background in biological sciences and I understand the notion that you cannot simply conclude on the basis of what might seem 'conventional wisdom', moreover, you try to disprove a theory before you prove a theory as correct.

I would be listening to LarryWill because I think he knows what he is talking about when he mentions that heat is not the same thing as temperature and how flow rate relates to heat removal. Most of the results of testing you see posted on the internet, while well intentioned and posted by people with genuine interest, should be taken with a grain of salt and seen as useful only for comparative purposes, not as absolute data. The tests are not double-blind and in the majority of cases do not provide error margins or statistical analyses and as such can be hardly considered 'scientific' in their approach. Fluid dynamics involves complex equations, you can't just add and subtract things from various test results like it was a shopping basket full of goodies and conclude on the basis of that.

Anyway, I'm sure I won't get away with impunity for my input but to help you with your question gobase I would suggest as Larry has by running two loops, one for your CPU/GPU combination and another for everything else.

Ahhhh....my friend cc01. I could have guessed you would want to stick your two cents in here. My comments....

Zitat von »cc01«

They are called 'laws' because they do not vary anywhere in the universe (except when conditions approach the speed of light) and have been derived/discovered through science and repeated experiment to test the hypotheses behind them.

Well....actually the affinity laws, or pump laws for simplicity, have several limitations on their use. Even more limitations when looking at options of adding multiple pumps to a loop (which is what we were talking about)...because the pump laws were intended to predict changes in a pump curve (for a single pump) if you change the impeller size or speed. They are useful for providing estimates and predictions if you already know the pump curve. You can't create a pump curve using the affinity laws, without having already measured a real pump curve. The affinity laws are useful for predicting the pump curve...based on a change in pump speed....if you already have measured the actual pump curve for a given speed. That is what I was copying for gobase....the actual pump curve. And the actual pump curves will be different for even the same pump....when a different pump top is used on it. You can't use the affinity laws to build a pump curve from scratch....only to predict how the pump curve changes when you change the speed or the impeller size. So for gobase....or 99% of the watercoolers out there...the most useful information to have is the actual pump curve.

Zitat von »cc01«

Although I am not a fluid dynamics expert I have a science background in biological sciences and I understand the notion that you cannot simply conclude on the basis of what might seem 'conventional wisdom', moreover, you try to disprove a theory before you prove a theory as correct.

I haven't "concluded on the basis of what might seem conventional wisdom".....all of my responses were based on the results of independent testing done in the water cooling community. I also haven't been discussing any "theories". Just the actual results of studies. There have been numerous independent studies in the water cooling community that simply measure the actual outcomes of temperature based on different flow rates. In almost all cases...they show lower temperatures for your components as flow rate increases. But only up to a point. Even LarryWill mentioned that his temperatures improved as his flow rate increased. He just doesn't feel the need for the 3 degrees lower temperatures he mentioned he got. The study I linked showed more like 5 degrees better as you increased flow....but everybody's system will be different. Everyone will also have their own goals for how their cooling system should perform. But most people who have chosen to tackle the additional expense of water cooling....are interested in performance.

Zitat von »cc01«

Most of the results of testing you see posted on the internet, while well intentioned and posted by people with genuine interest, should be taken with a grain of salt and seen as useful only for comparative purposes, not as absolute data. The tests are not double-blind.....

A double blind water cooling test? Look up the definition of "double blind" and you'll have as good a chuckle as I did at that one. Are you suggesting that the waterblocks themselves shouldn't be told if they are in a control group or not?

Zitat von »cc01«

Fluid dynamics involves complex equations, you can't just add and subtract things from various test results like it was a shopping basket full of goodies and conclude on the basis of that.

I went back and read all the posts to make sure....so I'm fairly confident when I say that "no one has done that". But if you are referring to the list of blocks and approximate pressure drops for those blocks....then yes...the total pressure drop should be estimated by taking the sum of the individual pressure drops. With caveats of course.


I've done the tit-for-tat with cc01 before...so I'm not going to get into that again. He seems to just like to argue with me no matter what I say. Larry and I were actually on the same page about most of the issues. We both suggested more than the one pump you were going to use. We also both agreed that increasing your flow rate would decrease the temperature of your components. The only thing we disagreed about really was whether a target of 1 gpm was excessive. And there is no equation or formula that can determine that for you. If I were you, I would just look at as many independent reviews as you can, and maybe post the question is several forums to get lots of opinions and experiences from some more actual water coolers. Also,...the number of blocks you are planning on using puts your loop into the category of very restrictive. You might see if you can find another build using 12+ blocks/rads and see what they are using for pumping power....and what kind of flow and/or temps it yielded. Again....good luck!