During Linpack tests inside a closed system case using the “weakest” cooling system of the today’s testing participants we managed to overclock our 45 nm quad-core processor to 3.68 GHz (+22.7%). The nominal processor Vcore was increased to ~1.46275 V in the mainboard BIOS (+27.2%).
During the OCCT tests in the same conditions the CPU remained stable up to 3.9 GHz (+30%) at 1.55V (+34.8%) Vcore.
Before we continue to our benchmark results, we would like to dwell on one very important thing. When the cooler was tested inside the system case installed the way you have just seen on the photo above (the airflow is directed towards the back of the case), Zalman CNPS9900 LED cooled our overclocked processor 2°C worse than CNPS9700! I was extremely frustrated with the obtained result and started looking for an explanation. As a result, I discovered that the cooler should be installed inside the ATX system case so that the wider part of its heatsink and two heatpipes entering it from the base were at the top:
After the events had taken this turn (in direct and indirect meaning of this word), Zalman CNPS9900 LED started to outperform its predecessor. It is in fact a very interesting observation, especially since the manual doesn’t say a word about it and in an open testbed when the cooler stands vertically we saw no difference in the efficiency depending on the cooler positioning. The thermal compound imprint on the processor heat-spreader was ideal in both cases and Zalman CNPS9700 demonstrated the same results in all positions. Looks like the direction of the main cooler heatpipes does affect the cooling efficiency in the end. However, I am going to get back to checking it out on Core i7 platform later on. So, let’s take a look at the obtained results.
The cooling efficiency readings are shown in the table and on the diagram below:
First of all I would like to say that the newcomer is definitely more efficient than its predecessor – Zalman CNPS9700 NT: the advantage is about 4~7°C depending on the testing conditions and type of CPU workload. Not bad, but it is still not enough to outperform Thermalright SI-128 SE. Zalman CNPS9900 LED proves as efficient as the latter only when Thermalright cooler fan works in its quiet mode. At higher fan rotation speed SI-128 SE is beyond any competition. Moreover, it turned out that Enermax Magma fan rotating at the same speed as Scythe Ultra Kaze fan keeps the CPU 2°C cooler. Moreover, further increase in the fan rotation speed (over 1600 RPM) doesn’t improve the CPU temperature any more. But, enough of Thermalright SI-128 SE discussion, let’s get back to our today’s hero - Zalman CNPS9900 LED.
What could we add here? There is one thing! You may have already noticed that the plastic housing covering the fan all around has cut-out slits in it. These slits release partially the airflow and I decided to get it to work for my benefit. I used regular clear scotch tape that appeared of the same width as the plastic housing. This simple modification improved the cooling efficiency in Linpack under maximum workload by 2°C! You may think it is a trifle, but this improvement can be achieved with the most primitive fix that will take you 30 seconds to put in place and will not change the looks of the cooler. Therefore, I would obviously like to ask the manufacturer in the first place: why did they make these slits at all, if they lower the cooler efficiency? Are they intended to improve the cooling of the around-the-socket area or to lower the noise? I doubt it. So, the question still stands.
Two best coolers in our today’s test session were also tested during maximum CPU overclocking. The test was performed in an open testbed with the workload created by Linpack. Here are the results:
As you see, SI-128 SE not only gets the CPU frequency about 50MHz higher than CNPS9900 LED, but also ensures lower operational temperature (84°C vs 92°C on the hottest CPU core).