While I generally agree with Mike, I am going to have to respond to some of these points as they are incorrect, at least when applied across the board.
MikeC wrote:
3. With heatsinks that have 10/12 "heatpipe paths"* into the fins, I think they actually move more heat more quickly than most WC systems -- except the most high end ones that use huge tubes, very large reservoirs (ie, a lot of water) and, naturally, very powerful pumps.
The size of the reservoir makes no difference. The amount of water in the system does not effect the cooling ability at all (it does obviously increase the amount of time to reach equilibrium, but does not effect the maximum heat load) -- okay technically with very large reservoirs with a large surface area it will, but in any sane sized system...
The size of the tubing makes a small difference at best, when we are talking about anything above 3/8" -- Cathar showed this a long time ago, the tubing size makes very little difference to flow -- again, technically using extreme sizes this is not true, but for the commonly used sizes, very little difference.
Flow rates above a certain rate (typically 1-1.5gpm) make very little difference. The item in the loop that is most effected by flow rates are the CPU/GPU blocks, and Skinnee's testing shows that there is massively diminished returns beyond 1-1.5gpm, and this is easily achievable with todays pumps, without needing the over sized pumps of yester-year.
MikeC wrote:
For cooling a single CPU in a PC, I think WC is completely outclassed by the best heatsinks. There's no contest for efficiency, cost, noise or even performance.
This statement.... what can I say? performance with water cooling FAR out classes air cooling. I'll give you cost and efficiency, I've said so myself above. Noise is somewhat debatable, as it depends on what you are trying to cool. Slow moving fans over a radiator are quieter than Delta's on a heatsink. That said, for a normal to low power system, yes, AC is quieter. But performance? no way. Not even close. The sheer surface area of a 140x3 radiator allows a much greater cooling ability than any AC heatsink -- I guess technically you could make an AC heatsink that large, but it would be impossible to manage. Water gives you flexibility that you cannot have with air.
MikeC wrote:
Where WC still has an edge is in cooling GPUs, mostly because of the sheer amount of heat the most powerful ones produce these days, and their awkward placement on a slot card. However, if a heatpipe/heatsink maker worked closely with a case maker, so that multiple heatpipes could be run from the GPU to fins mounted on the outside of the case like a water radiator, I'm sure an air-cooled heatsink could match a high end WC system -- without the noise of the pump.
Sure, a custom case with a massive passive heatsink may be able to keep up. But 2 (or more!) GTX 480s pump out nearly 600 watts... which is a huge heatload, and would require a very large passive heatsink. The guys with reservators typically max out their heatload at 300 watts ish.
Mike you have focused on the efficiency of heat pipes, but left out the surface area. While its possible that a heat pipe is more efficient than a modern WC block at heat transfer (and I'd question this anyway, modern water blocks are pretty good), the surface area available to WC is always realistically going to be much more than available for AC.
EDIT: Links for anyone who cares:
Effect of tubing size on performance:
http://www.xtremesystems.org/forums/sho ... p?t=147767
Effect of flow rate on CPU block efficiency:
http://skinneelabs.com/i7-blocks-2.html