Zalman CNPS8900 Extreme CPU Cooler Review
By Rafael Otto Coelho on April 26, 2012
The Zalman CNPS8900 Extreme is a low-profile CPU cooler. It has two heatpipes and a 110 mm fan. Let’s test it!
The beautiful box of the CNPS8900 Extreme has a big transparent window that allows you to see the entire cooler, as seen in Figure 1.
Figure 2 shows the contents of the box: the cooler itself, a small bag of thermal compound, a manual, a case sticker, and installation hardware.
Figure 3 displays the Zalman CNPS8900 Extreme.
This cooler is discussed in detail in the following pages.
Figure 4 illustrates the side (or front, since the cooler has a radial symmetry). Here you see the cooler is not too tall (2.36 inches or 60 mm), which is expected since the CNPS8900 Extreme is sold as a slim cooler.
Figure 5 reveals the front (or side) of the cooler, where you can see how the heatpipes come off the cooler base and turn around the heatsink. Keep in mind that the fins are made of aluminum, not copper as it seems. You can also notice that the fan connector is not the common type, but rather the type used on VGA coolers. The cooler comes with an adapter harness.
Viewed from the top, you see the transparent 110 mm fan, which doesn’t have any LEDs.
In Figure 7, you can see the bottom of the cooler. The heatpipes keep direct contact with the CPU.
Figure 8 reveals a closer view of the base, which doesn’t have a mirrored look.
The first step to install the CNPS8900 Extreme is to screw the clips on the base of the cooler. Figure 9 shows the clips for use with Intel CPUs installed. AMD systems use a similar pair of clips.
You also need to install the nuts on the backplate in the holes according to your CPU socket, as seen in Figure 10.
After those two steps, put the backplate on the solder side of the motherboard, put the cooler over the CPU, and fasten four screws.
We tested the cooler with a Core i5-2500K CPU (quad-core, 3.3 GHz), which is a socket LGA1155 processor with a 95 W TDP (Thermal Design Power). In order to get higher thermal dissipation, we overclocked it to 4.0 GHz (100 MHz base clock and x40 multiplier), with 1.3 V core voltage (Vcore). This CPU was able to reach 4.8 GHz with its default core voltage, but at this setting, the processor enters thermal throttling when using mainstream coolers, reducing the clock and thus the thermal dissipation. This could interfere with the temperature readings, so we chose to maintain a moderate overclocking.
We measured noise and temperature with the CPU under full load. In order to get 100% CPU usage in all cores, we ran Prime 95 25.11 with the “In-place Large FFTs” option. (In this version, the software uses all available threads.)
We compared the tested cooler to other coolers we already tested. Note that the results cannot be compared to measures taken on a different hardware configuration, so we retested some “old” coolers with this new methodology. This means you can find different values in older reviews than the values you will read on the next page. Every cooler was tested with the thermal compound that comes with it.
Room temperature measurements were taken with a digital thermometer. The core temperature was read with the SpeedFan program (available from the CPU thermal sensors), using an arithmetic average of the core temperature readings.
During the tests, the panels of the computer case were closed. The front and rear case fans were spinning at minimum speed in order to simulate the “normal” cooler use on a well-ventilated case. We assume that is the common setup used by a cooling enthusiast or overclocker.
The sound pressure level (SPL) was measured with a digital noise meter, with its sensor placed near the top opening of the case. This measurement is only for comparison purposes, because a precise SPL measurement needs to be made inside an acoustically insulated room with no other noise sources, which is not the case here.
Operating System Configuration
We adopted a 2°C error margin, meaning temperature differences below 2°C are considered irrelevant.
The table below presents the results of our measurements. We repeated the same test on all coolers listed below. Each measurement was taken with the CPU at full load. In the models with a fan supporting PWM, the motherboard controlled the fan speed according to core load and temperature. On coolers with an integrated fan controller, the fan was set at the full speed.
|Cooler||Room Temp.||Noise||Speed||Core Temp.||Temp. Diff.|
|Cooler Master Hyper TX3||18 °C||50 dBA||2850 rpm||69 °C||51 °C|
|Corsair A70||23 °C||51 dBA||2000 rpm||66 °C||43 °C|
|Corsair H100||26 °C||62 dBA||2000 rpm||64 °C||38 °C|
|EVGA Superclock||26 °C||57 dBA||2550 rpm||67 °C||41 °C|
|NZXT HAVIK 140||20 °C||46 dBA||1250 rpm||65 °C||45 °C|
|Thermalright True Spirit 120||26 °C||42 dBA||1500 rpm||82 °C||56 °C|
|Zalman CNPS12X||26 °C||43 dBA||1200 rpm||71 °C||45 °C|
|Zalman CNPS9900 Max||20 °C||51 dBA||1700 rpm||62 °C||42 °C|
|Titan Fenrir Siberia Edition||22 °C||50 dBA||2400 rpm||65 °C||43 °C|
|SilenX EFZ-120HA5||18 °C||44 dBA||1500 rpm||70 °C||52 °C|
|Noctua NH-L12||20 °C||44 dBA||1450 rpm||70 °C||50 °C|
|Zalman CNPS8900 Extreme||21 °C||53 dBA||2550 rpm||71 °C||50 °C|
In the graph below, you can see how many degrees Celsius hotter the CPU core is than the air outside the case. The lower this difference, the better is the performance of the cooler.
In the graph below, you can see how many decibels of noise each cooler makes.
The main specifications for the Zalman CNPS8900 Extreme CPU cooler include:
* Researched at Newegg.com on the day we published this review.
The Zalman CNPS8900 Extreme is a beautiful CPU cooler, with a low profile (2.36 inches or 60 mm) and good cooling performance for this category of cooler. It presented the same cooling performance as the Noctua NH-L12, which we recently tested.
The only problem is that the product website announces it as an “ultra quiet slim CPU cooler.” It is slim, but not quiet at all. Its 110 mm fan spinning at more than 2,500 rpm makes a high-pitched annoying noise, which makes it a bad choice for HTPCs or for any application where silence is desirable.