Testing

The thermal testing for the Thermaltake CL-P0191 will consist of monitoring temperatures with the system in various states, to be described in greater detail below.  We will also make note of the noise output after addressing the thermal results.

The thermal results were recorded from the motherboard's onboard thermal sensor for each state described in the list below:

1a. System Idle - CL-P0191 with thermal pad
1b. System Active - CL-P0191 with thermal pad
2a. System Idle - CL-P0191 with upgraded thermal paste
2b. System Active - CL-P0191 with upgraded thermal paste
3a. System Idle - CL-P0191 with upgraded thermal paste and alternate fan
3b. System Active - CL-P0191 with upgraded thermal paste and alternate fan

As you can see, we'll try the CL-P0191 as is straight from the box, then we'll try it with the thermal pad removed and replaced with some 'upgraded' thermal paste, and then finally we'll try it with an alternate fan.  The upgraded thermal paste wasn't anything fancy, but merely a well applied layer of white thermal paste from Thermaltake (considered an upgrade merely because I don't generally care for pre-applied thermal pads).  The alternate fan is another Thermaltake product, a 92mm Smart Case fan capable of running up to 3000 RPM.

Idle conditions simply involved having the system sit at the Window's desktop with nothing else running. Active conditions involved having the system run SiSoft Sandra's CPU benchmarks in a loop for thirty minutes by using the "Burn In Wizard" option.

The chart below summarizes the results recorded during the six states listed previously, while maintaining an ambient temperature of 20 degrees Celsius:

The whole 800 series of Pentium D processors runs extremely warm, and although I would prefer to see lower temperatures myself, I am well aware that this Pentium D 840 can run safely and reliably at the temperatures shown in the default setup!

What we see in the data presented in the chart above is that the cooler can do better with some help.  Simply removing the thermal pad for some thermal paste is a no-brainer, and the note on the Crazy PC site that I mentioned before is dead on.  I only used fairly generic white paste to drop active temperatures substanitally, but getting something like Arctic Silver paste might bring that number down a few more notches.

Adding the Smart Case fan in place of the stock fan did help greatly, but at the expense of silence.  The stock fan was never noticable, and although that may be a good thing in general, it did concern me.  The stock fan could regulate its speed thanks to the thermal sensor, but never once left the range of 700 - 1500 RPM, despite having a maximum rated speed of 2500 RPM, I don't know what to think other than this is a poor design or perhaps a defective unit.  As a test, I even connected the fan straight to 12V, no thermal sensor or PWM control, and it still wouldn't ramp up to full speed.

If noise isn't your concern, and aftermarket fan can obviously drop temperatures a good deal, and may be worth investigating.  Using a fan speed controller might help you find the perfect balance between noise output and cooling performance.