Discussion water cooling tube routing

[FordFreak2013]

I'm trying to decide how to route the tubing in an upgrade. With my current system in a Lian-li PC-75, it goes pump>rad>CPU>GPU>pump. Since I was able to mount the rad with the connectors up front right above the pump, it was fairly simple. Especially since the rad is a U-flow type. I am unable to mount the radiator that way in the new case. Plus the design of it requires one port to be the input for optimum performance. This is basically how the water flows through it.

I could have the water flow the same way which would cause more tubes to cross each other. Or I could change the path to pump>GPU>rad>CPU>pump.

This is what it looks like so far. If I change the route, I can rotate the CPU block 180. The barb is on the input for the block. I marked the in and out for the radiator for optimum performance. The GPU has spots for connectors on the top and bottom. I plan on running the input or output on the bottom since the card will be higher than the connections on the pump.

Do you think changing it to pump>GPU>rad>CPU>pump is a good idea? Any other ideas that you can think of?

 

[Gt403cyl]

The order makes no difference in cooling, as long as there is flow it’s fine.

 

[FordFreak2013]

The order makes no difference in cooling, as long as there is flow it’s fine.

Which do you think would look better?

 

[aigomorla]

shortest loop possible is always the best.

Always res before pump as well so the pump never runs dry, however it seems your running a res pump combo, so this rule is moot.

So again, go with the shortest loop possible.
Shortest length = less restrictions overall = more passes the water makes at components and radiator = more optimized cooling.
There is no such thing as "letting the water cool down in the radiator longer" as that is not how things with in thermodynamics when in application to fluid dynamics, especially over a system which is based off equilibrium.

Loop order only applies when you have a single massive heat source greater then 350W... like having 2 monster gpu's in SLI before a CPU, as then you will see a temp gradient when running a flow less then 2gpm.