[to shake things up - in reverse order this time]
I have no parts at all that lock me into any design at this point. Just this design that I've become "engineeringly" comfortable with. It's all in my head ...
(that's what she keeps telling me as well) 
when is pump head not a pump head?
Consider this, an electric motors sits on a table next to a 10' piece of linear gear. This linear gear poses no load on the motor while sitting next to it. Now the motor starts to press this linear gear straight up, every inch adding to a load presented to the motor from the weight of the linear gear it's pushed vertical. Eventually the motor will stall due to the inability to "lift" any more gear weight. (say 10' of linear gear) Now ... lets take this 10' section of linear gear, wrap it around a wheel and place the motor on it. The will still have deal with inertia [or the lack there of] but it will not stall. Given a few seconds or minutes it would reach the motors spin rate minus some amount due to friction. In a water scenario a pump head measurement it taken with a "linear gear". A open ended tube with which the pump will see how much fluid it can push vertical. At some point the pump will no longer be able lift any additional fluid weight. In a closed loop, it'd be taking advantage of the returning fluid’s weight to counter act what would be the "head" or pressure side of this picture. The pump will take a bit to get the water "spinning" in our loop but eventually it will spin at the pump's rate
minus some flow resistance. All “my” theory but to you follow my logic?
...but if you leave your computer on....
that was the one unknown in my thinking. How fast I could heat the basement? I've kept the idea alive for now because of how well
this technology has worked. Albeit on a much larger scale than my basement.
...handicap a loop(s) by under sizing any portion of the tubing...
hmmm I'm not really "reducing" the tubing size as much as I'm giving the pumps unrestricted access to supply. It’s a 3/8” cooling system. I just have 1/2” supply lines to the pumps. Why 3/8”? In my design I have 3 pumps & 3 loops worth of volume. I didn’t want to have to think about 1 1/2” lines from the warm rez -> radiator -> cool rez. (1/2” times 3 = 1.5” total volume)
I really dig those Laing pumps. From an engineering stand point they’re nearly perfect. (ok well that input line was a real Homer Simpson) and people here have already shown that a straight in supply with no restrictions (read 1/2 supply line) that these little beggars can really pump it real good .... [wait - write that down I think we have a song here -
pump, pump it jam, pump it up.... ]
oh yeah.
...NB and SB probably don't need that kind of cooling ...
On the eVGA reference i680 board they've (nVidia) have designed a heat pipe that runs from the SB to the NB and the NB has a fairly good sized fin area *and* it's own fan. Somebody thought it might be a good idea to cool it better then the plain-ole HS or HSF. Then again that hardware might just be eye candy or "wow" factor. As I do not have not gotten any hardware yet... I can't say but this is why I included it on that loop.
...you didn't understand my suggestion ....
No, I think I did ... I just might be under estimating the ability of water to absorb that much thermal energy. You know, I was the only one in high school physics that argued against the theory that warm water freezes quicker then cool water does due to the molecules being more mobile. I'm looking at it from an over all thermal energy level stand point of view. You still have to remove "X" amount of thermal energy to get the water to freeze. But what you're telling me, and please anyone jump in and choose your sides here, is that water is efficient enough at absorbing thermal energy that in our little thermal engines none of the individual parts (CPU, GPUS, Chipset) generate enough thermal wattage to raise the coolant temps appreciably in one, series looped, circuit? So - someone could take the radiator out of the loop and run the system for a short period of time before the coolant temps rose appreciably. Correct?
In my current design - guessing and stabbing in the dark, what would you guess the deltaT be between the two rez tanks assuming an average 50% load on the GPU's and CPU? (I realize there are a lot of variables - guess)
I can certainly grab a hold of the "point of diminishing returns" here. It's one of the reason's I'm not spec'in this system with a quad-core. Not much to do for a "double the cost" investment. What I'm learning here [and exactly why I wanted to engage this forum] is that - sure you can have a triple looped system, but at 2-3 times the cost I'll not see the requisite raise in performance. Or at least lower temps in any of the individual heat sources. I hope this is your point and that it has been realized in "real life" experiences. I do not think I've seen anyone do comparative tests of a single looped system to a system with individual loops for each one of the major heat sources.