Incorporating GPU into the water cooling system
Incorporating GPU into the water cooling system
Hi Guys
I just purchased the EKWB X360 kit and am considering adding a GPU to the loop. Is it too much to add another 420 radians as well?
I was thinking about the sequence Res>Pump>CPU>420rad>GPU>360rad>res.
Also, does the DDC 3.2 PWM provide enough head for 2 rads? If not, I was thinking of using the Thermaltake D5 res/pump combo so it would look like: Res>DDC>CPU>420rad>Res>D5>GPU>360rad>res
Another question is whether running 2 rads in parallel helps with cooling, since the flow rate would be reduced through those radians. But that would mean water passing through both the CPU and GPU before being cooled. Thoughts?
First of all, what CPU model do you have and are you considering any overclocking?
A 240/280mm radiator is suitable for a standard to moderate overclock on the CPU.
If you add a GPU to the loop, you'd need another 120mm radiator, making a 360mm radiator adequate for the entire system.
If your goal is to extract every last bit of performance from the CPU, I recommend using a 360mm radiator and placing it right before the GPU in line.
For optimal cooling, aim for the shortest possible run time—order doesn't matter much once the system stabilizes.
I usually follow a counterclockwise path on my loop: Res > Pump > Radiator > CPU > Radiator > GPU > back to Res.
I've tested a loop with four GTX 670 cards, MB block, and placed the CPU in a Res > GPU's > MB > CPU order, then set the temperature to 480, 360, and back to Res.
Running it forwards or backwards, my CPU temperatures only fluctuated slightly.
Radiators are among the least constrained parts in a loop, which means they don't require as much parallelization. GPU blocks are quite restrictive, which explains why most users run them in parallel with more than two instances.
adding as much rad surface area as possible isn't excessive, since it helps maintain fans at lower RPM, which keeps the system stable. DDC 3.2 can manage this loop effectively. The key points about loop order are:
* simplify the design
* prevent vertical zigzags
* ensure a clear fill port at the top and a drain port at the bottom
component arrangement doesn't matter much because the temperature difference across the loop stays within 1°C.
tuning radiators in parallel isn't ideal. The longer the liquid stays in the radiator, the colder it becomes, making cooling harder (smaller temperature gap means slower heat transfer). So performance remains similar. However, with reduced flow and parallelism, bleeding air from the radiators becomes very difficult.
i wouldn't suggest running multiple components in parallel when using a multi-GPU setup, even though the loop includes both CPU and GPU in parallel. This is mainly for appearance, but it reduces practicality in maintenance and troubleshooting. For instance, I once struggled to identify a clogged CPU block caused by EK's anti-vortex foam that had partially dissolved and broken apart. It took time until the foam was sufficiently broken down before I could detect the issue, which happened long after the CPU block was already affected. You really need to inspect the CPU block directly to see if microchannels are blocked.