Probably because it doesn't matter. Anytime you test a block you have to deal with measurement errors. The most obvious ones involve temperature, but you also have power measurements (for the CPU, pump, and all the other stuff generating heat), flow measurements, etc.
In terms of a block's performance, you really only need to know an inlet temperature (or outlet temperature) and flowrate to go along with CPU wattage. Fluid properties dictate, and I really mean DICTATE how much the fluid temperature will change. So long as your system is operating at steady-state, who cares if the water comes from an "infinite reservoir" or gets recirculated, it's still entering the block at a known condition (within our measurement error, that is).
It really is not that complicated to conduct a test. The real trick lies in how well you can control and measure your test conditions. Drawing fluid from a very, very large reservoir offers no inherent advantage over recirculating fluid once you reach steady-state. And make no mistake, even when drawing from a large reservoir you STILL need to wait for the block to reach a steady-state condition.
Personally I prefer an all inclusive test of a real world system. Unfortunately, the range of complete kits available isn't very extensive so we're left cobbling together bits and pieces from multiple sources. This makes gaging performance before assembly is complete rather a difficult task. Yeah, individual reviews can help in this area, but you still won't know how everything will work together as a group until you actually put it together and run it.
|