waterblock/plumbing performance metric

[af4now]

Here’s a practice that I plan to use to benchmark the effective performance of my waterblock and flow (pump and tubing). It cuts a lot of the environmental and system noise that can affect benchmarking with CPU temp alone and helps isolate some of the hot topics discussed on [H] forums. This is only a partial performance measure because I don’t believe there’s a simple and accurate way to include the effectiveness of the radiator; however, 99% of the fuss on this forum is about the flow restriction of waterblocks and tubing anyway.

The thermal resistance (1/conductivity) of your system is a good measure of cooling performance and it is something that can be reasonably measured on a macro scale if you know the heat flux of the processor in [watts]. If you don’t know the heat flux (CPU power consumption is a close estimate), you can still compare like processors running identical loads.

Heat energy always flows from hot to cold and the job of your cooling contraption is to provide a medium for that heat transfer. The physical properties are analogous to that of an electrical system:

CPU Temp . . . . . .Thermal Resistance . . . . . .Water Temp
. . . . . . .------------(power flow in watts)-------------->

. . O-----------------------/\/\/\/\/\/\/\/\/\/\/\/\/--------------------O

. . . . . . . ------------(power flow in watts)-------------->
Voltage A . . . . . . Electrical Resistance . . . . . .Voltage B

The closer the temperature of the water is to that of your CPU temp, the better the performance of your cooling system. In an ideal system, there would be zero thermal resistance so that the water temp would be identical to that of the CPU; analogous to Voltage A being equal to Voltage B if there was no electrical resistance.

This, of course, can only be done once the thermal system has reached steady state - incuding the CPU load.

Things that will be measured are the following:
- how well the block is bonded to the die with thermal compound **
- effectiveness of your particular waterblock as a heat exchanger **
- flow rate effects (pump, tubing, and block flow impedance)
- turbulent vs laminar effects
- thermal properties of the fluid

** I believe these are underrated factors and are otherwise hard to measure

 

[Top Nurse]

So when and how?

 

[af4now]

How:

a) You will need to get a measurement of the temperature of your processor and of your water.

subtract Temp_CPU - Temp_water : delta_T in [deg C]

b) You'll need to calculate the power output of your processor:

get CPU Power from Here and install

get stats for your processor from Here

Enter the stats of your processor and enter your overclock and it gives you a result: Q in [watts].

c)

thermal conductivity = Q / delta_T
thermal resistance = delta_T / Q

This isn’t the rocket surgeon answer, but this number will be a fairly good estimate of the relative effectiveness of the thermal compound -> water block -> plumbing of the cooling system. It is likely that thermal resistance will vary at different temperatures and power levels.



When:
June 6th for me, can't do anything before then