Coolant/Fluid Roundup – Thermal Performance

Thermal Testing Methodology/Specification

If you are a regular Skinnee Labs reader then you already know we do things a little differently. For those of you who found your way here for the first time, strap yourself in because we go all out on our test methodology and specification. We are not here for publishing fluff pieces, we want the hard data on every product we test even if that means more work. After all, we want to know how well each product performs and where to spend our money as well… we are enthusiasts just like you are. With that said, let’s cover the test methodology for coolant testing.


Starting from an empty loop, the top of the reservoir is removed and filled with fluid as gravity pushes the fluid out both ports of the reservoir. I continually add fluid as the level in the reservoir drops. Once there is enough fluid in the loop, the pump is started using a secondary power supply and shutoff once the fluid level is low enough for the pump to suck in air. I repeat the filling and pump cycling process until the loop is full, where I then top off the reservoir to the point where I can screw on the cap without fluid leaking out the top. I then move the molex from the secondary power supply back to the molex lead on the system power supply and power on the system. Once the system has booted into Windows, OCCT, GPU-Z 0.4.3, Wintest and MSI Kombustor 1.0.7 Multi-GPU are launched and the six-hour test run is started. BIOS controls the CPU overclock parameters and MSI Afterburner 1.6.0b6 holds the GPU clock profile which is automatically applied once Windows loads.

The first two hours of the run are considered warm-up or time for the loop to reach equilibrium under load, the remaining four hours are the important part. The four hours after warm-up are the data used for calculation, which you will see for each fluid on the results page. Once the full six-hour run was complete, each of the test applications was closed and re-opened, affinity was set for Kombustor Multi-GPU to all cores, GPU-Z and WinTest left. The test was not started immediately, I waited a minimum of thirty minutes for the loop to reach an idle state before kicking off the next run. After the cool down period, GPU-Z and WinTest set to log, OCCT was started first and then Kombustor for another six hours.

After three runs, the rig was set into hibernation and the loop was drained back into the original container or mix container in the case of Hydrx. However, some fluid would have remained. The radiator to reservoir tube was disconnected and lung power was used to push all the fluid out of the reservoir into the drain line. After the radiator was fully drained I did the same to the pump and block side of the loop, again using lung power to blow out as much fluid as possible. The radiator to reservoir tube was reconnected, drain valve closed and the loop filled with tap water with a full bleed process. The drain process repeated clearing the loop of tap water. However, a minimum of three tap water flushes was performed between changing coolants to ensure there was no cross contamination. Even with the lung power assisted drain, a very small amount (10-15mL) of tap water remained since a zero-mount rule was followed through testing. Considering the loop held 900mL, 10-15mL provided no advantage or disadvantage to any fluid in the roundup.

Now that the loop is cleaned with a miniscule amount of tap water refusing to leave, the drain valve was closed and the loop filled and bled using the next coolant. To be clear, a zero mount rule was followed during testing where the GTZ was never removed and the GPU’s stayed in their cozy PCI-e slot with their blocks on.


There is quite a bit of crossover/overlap from the Methodology and Specification sections, but the specification covers all the nitty-gritty details from the overall methodology.

You can see the loop order in one of the photos on the bottom of the page, but just to be clearly state all components, the loop order for all tests consisted of Reservoir -> Pump -> CPU Block -> GPU Block 1 -> GPU Block 2 -> Flow Meter -> Delrin T (drain) -> Delrin T (Temp1) -> Delrin T (Temp2) -> Radiator -> Delrin T (Temp3) -> Delrin T (Temp4) -> Reservoir.

Below is list of all of the tools, gadgets, goop along with all the hardware used for testing.

  • Hardware Platform:
  • Case: MSI Big Bang display bench
  • Cooling Loop:
    • Radiator – XSPC RX360 V2
    • Fans – 3x Yate Loon D12SL-12’s
    • Pump – Swiftech MCP355/Laing DDC3.2 with XSPC V3 Top
    • CPU Block – Swiftech Apogee GTZ
    • GPU Blocks – Danger Den Tieton GTX280 w/ RAM cooling backplate
    • Drain –Delrin T with Alphacool G1/4 Valve
    • Reservoir – EK Multi-res 150, multi option bottom, left inlet from radiator/right outlet to pump
    • Flow Meter –Koolance FM17 (calibrated to King Instruments 7520)
    • Delrin T sensors –Thermochill and DangerDen Delrin T’s with BP G1/4 fittings housing 1 probe each
    • Barbs – DangerDen/Bitspower 1/2″ Fatboys
  • Thermal Interface Material: Arctic Cooling MX-2 on CPU and GPU’s, GPU’s used DangerDen stock thermal pads on VRM’s and memory on back of PCB
  • Temperature Monitoring and Logging: CrystalFontz CFA-635 with SCAB attachment – Used to log 16 temperature sensors at 1 second intervals for the full 6 hour duration using WinTest.
  • Thermal Sensors: Dallas DS18B20 Digital one-wire sensors – .5C absolute accuracy overall with a .2C mean error between 20-30C.
  • Test Bench Sensors Deployed:
    • 6 Radiator Air In sensors
    • 6 Radiator Air Out sensors
    • 2 Radiator In sensors
    • 2 Radiator Out sensors
  • Temperature Sensor Data: Radiator is equipped with 6 Air In temp sensors and 6 Air Out sensors (2 per fan) logged every 1 second for the duration of the test. Both Air In and Air Out sensors are averaged over the duration of the test, which provides 21600 data points per sensor for each test.
  • CPU/GPU Core Temperature Monitoring and Logging: OCCT
    • CPU core temperatures are logged every 1 second, providing 21600 data points per core.
    • GPU (non-driver) for each card are logged every 1 second, providing 21600 data points per card.
  • GPU Load: MSI Kombustor v1.0.7 set for 1280×1024, 4X MSAA
  • Data Logging: Each temperature sensor and fan RPM channel is logged for 6 hours, 2-hour warm-up and 4 hours of compiled data. OCCT, Kombustor, WinTest and GPU-Z are running for the entire test duration, with OCCT, WinTest and GPU-Z logging each sensor/data point every 1 second for the entire 6-hour run.
  • Test Lab Environment: Unfortunately, I do not have an environmental test chamber. All tests are performed in 10×13 room in my basement (aka: The Lab) which is temperature controlled via a wall thermostat and on a separate zone from the rest of the house. I am able to maintain a somewhat consistent room temperature this way. However, the room does have some temperature variance.

The shots above were snapped as Feser One blue was in the loop, I just could not resist taking the photos since the fluid went so well with the color scheme of the Gigabyte UD7. Plus, this gives the visual of the loop details and descriptions stated above in the specification.

There you have it, the full test methodology and specification. Without further delay, let us move on to the performance results.

Pages: 1 2 3 4

Leave a Reply

You must be logged in to post a comment.