by Sergey Lepilov
07/20/2006 | 10:06 PM
While Intel and AMD are both working hard to reduce power consumption of their existing and upcoming processors, it would be unreasonable not to use the advances in heatsink design and heat-conductive materials that have been achieved in the research for highest-efficiency cooling. Having efficient heatsinks on one hand, and CPUs with low power dissipation on the other, it is now possible for the manufacturers to try to make their coolers smaller or reduce the noise they produce. It’s in this direction that two world-famous brands, Zalman and Scythe, have been moving recently.
In this article we will introduce to you Zalman CNPS8000 and Scythe Mine (SCMN-1000) coolers and compare their performance with some already tested models. We will also check how much noise they make, how easy they are to install, and how their prices compare.
Cooling systems from the Korea-headquartered Zalman have long and deservedly won renown in the overclocking community. Besides boasting high efficiency, Zalman-branded devices for cooling nearly every system component are also always up to the highest aesthetic standards. It’s just pleasing for the eye to look at the design and package of most of Zalman’s products. The new CPU cooler I am going to test today – Zalman CNPS8000 – is not an exception.
Here’s the package the cooler comes in. It is a small dark-blue box with a carry handle:
There are two transparent windows on the face and reverse sides of the package so that you could take a look at the cooler before purchasing it. The main features of the Zalman CNPS8000 are listed on the face side; this information is completed with a detailed specification at the back of the package. The cooler is manufactured in Korea.
Inside this cardboard box there is a plastic packet with the cooler and accessories:
And this is a photo of the cooler itself:
Four copper 6mm heat pipes marked as “TT1586” have contact with a copper base about 8mm thick. The cooler’s heatsink is put on the heat pipes. It consists of 18 low-profile plates on the two bottommost heat pipes while the 44 high-profile aluminum plates in the cooler’s central part have contact with all the four pipes and are shaped like a cup. The plates are about 0.25-0.3mm thick. The area of contact of the heatsink’s ribs with the heat pipes is enlarged by means of bottlenecks I have noticed before in other heat pipes based coolers. The whole arrangement is topped by a plastic frame with a 7-blade 85mm fan that runs on two ball bearings. You can use the Fan Mate 2 controller to vary the fan speed from 1400rpm to 2800rpm. If you attach the fan directly to the mainboard, its speed will be 3000rpm.
The heat pipes lie in the grooves in the copper base of the cooler.
Advertising its CNPS8000, Zalman draws your attention not only to the option of using this cooler in Ultra Quiet mode, but also to its low height which is a mere 62.5mm. Zalman’s specialists think that with these dimensions the cooler can be used in compact system cases and in barebone systems.
The cooler is 108mm wide, and it needs 54 millimeters of free space around the CPU socket, counting from the heatsink’s center, to mount normally on the mainboard.
Obviously, an overwhelming majority of mainboards meet this requirement. It’s not that simple with the cooler’s length, however.
It’s all well where the heat pipes end – you need a gap of only 50 millimeters. But you need 58 millimeters on the side where the pipes are curved. The instructions enclosed with the cooler contain an error as they specify dimensions of 50 millimeters for both sides, but one side is in fact longer by nearly a centimeter (this is confirmed in the 1.6MB, PDF manual you can find on the official website).
This may seem like a small thing, but it was the curved part of the heat pipes that prevented me from installing the Zalman CNPS8000 on an ABIT AN8 SLI mainboard (nForce4 SLI, Socket 939) in one of the two possible positions because the memory modules in Slots 1 and 2 got in the cooler’s way. It’s of course no problem to move the memory into Slots 3 and 4, but the AMD Athlon 64’s memory controller is known to be vary capricious about fine settings of memory timings which differ for different slots (this is true for nForce4-based platforms, at least) and putting your memory into Slots 3 and 4 isn’t the best solution if you want to have maximum performance.
Platforms with the LGA775 socket offer twice the freedom in choosing the cooler orientation, but you may find that capacitors or heatsinks near the socket restrict your choice. For example, when we took an ABIT AA8 DuraMAX mainboard (i925XE, LGA775), it was the chipset’s heatsink that acted as an obstacle to installing the CNPS8000 in one of the possible positions. So, if you are going to choose this cooler, make sure nothing will prevent its installation on your particular mainboard.
The cooler’s base is ideally flat and finished, but not to a mirror shine.
The Zalman CNPS800 can be installed on LGA775 and Socket 754, 939 and 940. I guess it is not surprising anymore that modern coolers do not support Socket A (462), but it’s the first time I see a cooler that is not compatible with Socket 478. This is quite logical, though, since older sockets get less attention nowadays, so I don’t count the lack of support for Socket 478 among the CNPS8000’s drawbacks. The new Socket AM2 is not on the list, either, but I dare suppose the cooler is compatible with it.
Except for the above-mentioned problems with the dimensions, it is in fact very easy to install the Zalman CNPS8000 on Socket 939. You don’t need any tools. Just attach the clips to the cooler’s base and hitch it to the standard plastic frame of the CPU socket.
The whole procedure is going to take a mere 2-3 minutes of your time, even counting in the time you’ll unpack the cooler. But if you want to remove the CNPS8000 from the processor, you’ll have to take a wide flat screwdriver and press with it down on a special jut in the fastening.
It’s all even simpler with LGA775, but longer because you have to take the mainboard out of the system case to fasten a special back-plate to it. Then you use the enclosed LGA775 fastener and screw it to the back-plate with a cross-head screwdriver.
You’ll learn about the specification and efficiency of the Zalman CNPS8000 in the next sections of the review. Right now let’s take a look at a cooler that came to our labs from Japan. It’s called Scythe Mine.
I personally am looking forward to testing the Scythe Infinity cooler the Japan-headquartered Scythe announced at Computex 2006. It seems to be an unprecedented air cooling solution with tremendous potential. The cooler I’m going to test today – the Scythe Mine – is no less interesting, however, as it claims to be a lucky combination of efficiency and low noise level. This is how it is introduced in the company’s press release. Let’s check if this claim is grounded well.
The cooler’s package is large and has a plastic window in the middle through which you can see most of the device.
Scythe made good use of the available package space, putting a lot of text on it. You can learn about the technologies implemented in the new cooler and its specification, check out the list of included accessories and read the warranty obligations.
The Japanese word “Mine” means a mountain peak and the heatsink’s plates are shaped accordingly:
The cooler represents the nearly-classic tower-like design with thin aluminum plates put on heat pipes. In the Scythe Mine there are two groups of plates between which a 100mm fan is located.
There are 2 x 46 plates in total and their thickness and stiffness is similar to that of the Zalman CNPS8000, i.e. 0.25-0.3mm. The cooler has only three heat pipes, 6mm in diameter, and this looks like a drawback to me. The heat pipes go through the cooler’s base and are sealed in their top with screw nuts.
When I first saw about this cooler I immediately had a question why they hadn’t cut those pipes into six and put their ends into the base? This would have increased the area of contact of the heat pipes with the copper plate in the base while keeping the manufacturing cost of the product roughly at the same level.
On one of the cooler’s sides arrows show the direction of the fan rotation and airflow, thus helping you orient the cooler in the right way inside the system case.
The new universal fastening system (Versatile Tool-Free Multiplatform System) can also be viewed in this photograph (for more detail on this fastening system see our article called Roundup: Seven High-Performance Coolers for AMD and Intel Processors).
The standard 10cm fan can be replaced with another fan with a diameter from 6 to 14cm. You only need to unfasten the two screws in the bottom part of the cooler. There is only one reservation: the width of the fan must be 25 millimeters. Thus, the Scythe Mine offers you the choice between silence (if you put in a quiet 6cm fan) and efficiency (a 12-14cm fan), but you have to purchase the fan separately in either case.
The cooler’s base is protected against scratches and damage with a strip of polyethylene film:
The cooler’s base is perfectly finished:
You can see the thermal paste and the camera as well as the window frame and clouds reflected in the cooler’s base.
No matter how well finished it is, the base of the Scythe Mine seems too thin to me at only 2mm. This may not be enough for efficient transfer of heat from the CPU heat-spreader to the heat pipes. This relatively thin copper plate with high thermal conductivity will quickly take heat off the CPU but will have nothing to distribute this heat in. The aluminum fastening/heatsink in the bottom part of the cooler contacts with the plate mostly through the heat pipes. Moreover, the pipes just lie on the copper plate, i.e. contact with it with their bottom part only. This fact, and the results of the tests, suggests that the cooler may be improved by increasing the number of heat pipes’ ends in its base and by making the base thicker.
The Scythe Mine’s standard 100mm fan is manufactured in China and rotates at 1500rpm. It consumes about 1.44 watts of power (12V x 0.12A).
Frankly speaking, I was expecting to find the silent version of the fan on Sony’s bearings, but I was wrong. Anyway, the fan is very, very quiet and the Advanced Wave Stack Fin technology implemented in the Scythe Mine makes it an almost silent cooler.
The cooler package contains the following:
Like the Zalman CNPS8000, the Scythe Mine is not compatible with Socket A (462), but contrary to the Zalman, supports Socket 478. I also don’t think there should be any problems with mounting the Scythe Mine on Socket AM2.
There’s not much sense in describing the installation procedure considering the simplicity of the Versatile Tool-Free Multiplatform System implemented in this cooler. You don’t need any tools, just your own hands. I found it not very convenient to install and uninstall the Scythe Mine on an LGA775 mainboard inside the system case because it was not handy to press on the clips and turn them. The heatsink’s ribs are located too close and there’s just too little room for your hands. It’s all much simpler if you take the mainboard out and then mount the cooler.
So, this is the end of the descriptive part of this review. I guess you’ve got some general notion about the new coolers from Zalman and Scythe, now you may want to have a look at the specifications of the coolers to be tested.
The table lists the official specifications of the coolers that will take part in today’s tests.
* - maximum speed when connected directly to the mainboard (without Fan Mate 2)
Besides the coolers listed in the table, we will also include the Pentagram Freezone NXC-100 Cu copper cooler.
For more details on our testing participants please see the corresponding articles called Battle of the Titans: Super Coolers from Scythe, Thermaltake and Zalman Face to Face, Two Towers: Noctua Heatpipe Coolers Review and Titan Vanessa S & L-Type and Thermaltake Big Typhoon vs. Zalman CNPS9500 LED.
We tested the coolers using the following hardware parts:
The tests were performed in Windows XP Professional Edition Service Pack 2. We used the same thermal paste with all the coolers and disabled the automatic fan speed management systems in the mainboards’ BIOSes.
SpeedFan version 4.28 was used to monitor the temperatures and fan speeds:
The CPU was heated up by running the FPU test from S&M version 1.8.0 (alpha) at 100% load for 15 minutes.
I also loaded the processors with SuperPI mod.1.5 by calculating pi to 32 million decimal places; this task takes a little less than 29 minutes on the described testbed. I guess this is enough to evaluate a cooler’s efficiency in a closed system case as well as on an open testbed. The temperature was read from the sensor integrated into the CPU. As usual, we control the thermal throttling of the Prescott-core Intel Pentium 4 processor by means of ThrottleWatch 2.02.
The coolers were tested under identical conditions, in a closed system case and on an open testbed. The system case was put into the niche of a computer desk and its 120mm system fan on a side panel was disabled. So, the cooling conditions in the system case were somewhat worse in comparison with our earlier tests. The room temperature remained constant at 25-26°C.
The tower-like coolers were positioned vertically on the open testbed and horizontally in the system case. Except for the Noctua NH-U12 on the AMD platform, all the coolers were installed in such a way that their airflow was directed towards the rear panel of the system case where a 120mm fan was located. The Zalman CNPS8000 was installed with the ends of its heat pipes towards the rear panel of the case, too.
To make the trial harder, I overclocked the AMD Athlon 64 3200+ processor from its default 2000MHz to 2700MHz, increasing its voltage to 1.6V.
Let’s see how the coolers do on an open testbed. In the top of the diagram the results of the coolers with their stock fans in quiet mode and at max fan speed are given (there is only one test mode for the Scythe Mine due to obvious reasons). In the middle part of the diagram there are results of three super-coolers and of the Scythe Mine all with the same 120mm fan in the quiet mode (~1450rpm). In the bottom part there are the same four coolers at the maximum fan speed (~2000rpm).
As you can see, the Zalman CNPS8000 in the quiet mode and the Scythe Mine deliver a rather mediocre performance in comparison with the super-coolers, but the Scythe Mine is 5°C better than the new cooler from Zalman. The all-copper Pentagram Freezone NXC-100 Cu is competitive against the heat pipes based CNPS8000 in the quiet mode, but costs much less. The Zalman leaves the Scythe Mine behind when its fan speed is increased, but becomes rather loud at that. Don’t forget, however, that you can replace the 100mm fan of the Scythe Mine with, say, a 120mm one. This simple operation helps the Scythe Mine outperform the Zalman CNPS8000. I’m talking about the quiet mode now, the 120mm fan rotating at about 1450rpm.
Increasing the fan speed further doesn’t affect the performance of the super-coolers on the open testbed. The Scythe Mine is the only one that lowered the CPU temperature by 2°C in comparison with the quiet mode.
Now let’s see what we have in a closed system case:
A nasty surprise, the system with the Zalman CNPS8000 in the quiet mode couldn’t pass the test. 73°C was the last mark on the monitoring graph. So, the new Zalman CNPS8000 is not suitable for cooling well-overclocked CPUs in a closed system case and in hot weather. That’s sad as I had expected better results from it. Yes, at the higher fan speed this cooler will do for overclocked single-core AMD processors with increased voltage and it matches the cooling efficiency of the Scythe Mine, but look at the results of the Pentagram Freezone NXC-100 Cu! It leaves no chance to its two opponents at the max fan speed and is as effective as the Scythe Mine in the quiet mode.
In its turn, the efficiency of the Scythe Mine can be improved to the level of the Scythe Ninja, the best super-cooler in this review, by simply replacing its stock 100mm fan with a 120mm one. That’s impressive, but Scythe unfortunately doesn’t ship its new cooler with a 120mm fan by default. All the four coolers do considerably better when their fan speed is increased, but I don’t think you will want to exchange the silence for the CPU temperature reduction of only 3°C.
As a final touch to my tests on the AMD platform, I decided to check the supposition that the Scythe Mine’s airflow was used in a non-effective way. The sides of this cooler are open, contrary to the Noctua coolers, for example. So, I sealed the sides, top and bottom of the Scythe Mine’s heatsink with a piece of scotch tape, expecting to see a considerable boost in its performance, but there was no miracle. The CPU temperature became lower by only 1°C!
I first wanted to test the coolers on a dual-core Intel Pentium D 820 that we managed to get, but the ABIT AA8 mainboard does not support dual-core modifications of the Intel Pentium 4 at all (I guess such processors are not supported by the i925XE chipset). So, I had to limit myself to an Intel Pentium 4 506 on an E0 stepping core (2666MHz, 533MHz FSB) which did well at overclocking, being stable at 3800MHz:
I didn’t even have to increase the core voltage to achieve that frequency (it is 1.385V by default). I should confess the processor didn’t react at all to my changing its voltage which was rather strange for a Prescott core. The mainboard’s limit is 244MHz which is a long way yet from the achieved 190MHz, and the memory frequency and timings were set up in such a way as not to restrain CPU overclocking. However, I had to stop at the frequency shown in the screenshot above.
Let’s first have a look at the results I got on an open testbed:
After doing very well in comparison with the Zalman CNPS8000 on the AMD platform, the Scythe Mine fails in this test. I thought that poor contact between the cooler’s base and the CPU heat-spreader might be the reason for this poor performance, but the results didn’t change after I made sure the thermal paste was applied properly and even tried different positions of the cooler’s heatsink and fan. Alas, the Scythe Mine with its stock 100mm fan can only compete with the much cheaper Pentagram Freezone NXC-100 Cu on the Intel platform.
The Zalman CNPS8000 in its turn looks very well on the Intel processor: the temperature of the single-core Intel Pentium 4 at 3800MHz clock rate and under the S&M load is no higher than 49°C even in the quiet fan mode (according to the sensor in the CPU itself). The temperature goes down by 2°C when the fan speed is increased, but I don’t think it’s worth the lost silence.
Now let’s see how these results will change in a closed system case:
It’s no different with the headliners of this review: the CPU temperature is just a little higher with the Zalman CNPS8000 and the Scythe Mine. There are, however, changes in the ranks of the super-coolers with the Thermaltake Big Typhoon suddenly on top, even though not by much relative to its opponents. Somewhat puzzled at this fact, I decided to build a diagram basing on the reading of the mainboard’s PWM sensor located near the CPU socket.
Quite expectedly the coolers that direct their airflow towards the mainboard beat their opponents that have a tower-like design and a horizontal airflow. This affected the result of the Thermaltake Big Typhoon, of course. But the Pentagram NXC-100 Cu and the Zalman CNPS8000 aren’t brilliant at cooling the overclocked processor, having lower overall performance.
To expand the topic of airflows, I want to show you a couple of interesting diagrams that show the CPU temperature graphs by the following sensors: CPU (red), mainboard’s near-socket PWM sensor (orange), and chipset (violet). The Scythe Ninja is first:
And now the Thermaltake Big Typhoon:
I had to group the temperature graphs due to the image width limitations. These results were obtained in a system case and with the same 120mm ~1450rpm fan used with both the coolers. This test was performed on an Intel Celeron D 351 3.2GHz overclocked to 4.1GHz with a voltage increase by 0.1V.
Easy to see, the Thermaltake Big Typhoon not only cools the CPU better, but is also more effective at cooling the mainboard near the CPU socket, which is quite an important thing. It is crucial that the mainboard’s power elements are cooled properly if a dual-core processor is used. That’s why coolers that direct their airflow towards the mainboard look preferable as ensuring higher overall stability of the system. I want to note that the Scythe Ninja’s fan was attached a little higher than the bottommost plates of the heatsink because the memory modules prevented me from placing it lower. If the fan is set lower, the temperature reading of the PWM sensor is going to be lower, too.
I think that the market success of the new coolers Zalman CNPS8000 and Scythe Mine will entirely depend on their price. Zalman CNPS8000 started selling for around $55-$60, but this is still rather expensive for a cooler that does not impress with its performance. Owners of compact system cases who are sure there’ll be no obstacles for the cooler near the mainboard’s CPU socket may like this model, though.
The Japanese “sandwich” Scythe Mine can be bought for about $43, i.e. only $3-4 less than the price of the super-cooler Scythe Ninja. The Ninja sells without a fan, so you should add $10-12 more to its price, but the Scythe Mine shows its full potential only with a 120mm fan which has to be purchased separately, too. So I don’t think your choice is obvious considering these prices. The Scythe Mine is a quiet cooler with good enough efficiency and simple fastening, but nothing more than that.
The pros and cons of the tested coolers are listed in brief below.
Scythe Mine (SCMN-1000)