The purpose of the prototype is to test the behaviour of the Submer bio-friendly coolant with a real 1U server at different temperatures and CPU loads. The prototype is composed of:
We're only focusing on modifying the server load, water circuit and coolant circuit temperatures to see the behaviour of the server and its CPU values. There's no focus on PUE (Power Usage Effectiveness) as it's very difficult to achieve reasonable measurements with such a minimal setup.
For now, we've done most of the testing based on an inlet coolant temperature of 35ºC (bottom of the plastic casing) and an outlet temperature of 45ºC (top of the casing). This is basically controlled by regulating the water and cooling pump speeds and the water chiller temperature setting. This setup has been keeping our CPU at 65ºC.
We'll be running more tests, specially on extreme temperature conditions and simulating worst case scenarios. As soon as we have extended data, we'll also publish it.
As time goes by, we'll publish additional posts to explain if we see any changes in behaviours of the coolant or the server materials.
Check out the video of the Phase 0 prototype: