A Roundup of High-end SocketA Coolers. Q1-2`2003

Introduction

Quite recently we tested a number of cheap (low-end) coolers for powerful (Barton 2000 MHz) processors. In that roundup we focused mainly on the cooler's ability to cool the processor at nominal clock speeds. In today's roundup we'll be looking into higher-end coolers while keeping the possible overclocking in mind (these are CoolerMaster Aero7 and Thermaltake Volcano 11). Along with these, we'll be testing coolers designed for maximum low-noise operation (Zalman 6000Cu and Igloo SilentBreeze).

Prior to tests, let me answer a couple of typical questions from readers of the previous roundup:

1) Why do we have to save on a cooler if we buy a high-end processor?
If a system is not meant for further overclocking at all, or if no upgrade is planned for the coming year or two, the most cheap cooler capable of doing a decent cooling job for the processor will suffice. Besides, AMD processors are getting cheaper quite fast, so a recent high-end processor turns mainstream very soon.

2) Why hasn't this or that cooler fallen within the testing roundup (among the most frequently mentioned coolers were Igloo models :).
Frankly, the division of coolers into cheap/expensive (low/high-end) is pretty conditional: you see, some regard high-end models as those starting with $10, for others they start with somewhere at $17. And of course the cooling kit varies. To bind the results of both roundups, we included results for the Titan TTC-CU5TB cooler which showed the best efficiency among the cheap (in my view) models.

Note that this review is one of the latest roundups of SocketA coolers. The thing is the Athlon XP line has come to its completion - there's been released presumably the last XP3200+ processor built on the Barton core featuring 200(400DDR) MHz bus. For the tests, I was able to acquire such a processor whose maximum overclocking upon the voltage raise amounted to ~2450 MHz. This means the clock speeds of those made following the 0.13 mk process technology reached their technological limit. So the only thing what AMD has in stock is to produce at best one or two models of processors not exceeding 2.3 GHz. The probability of that is small enough - in 2 months, Athlon 64 are to hit the retail, and manufacturers of cooling devices are already clustering in this market sector (a typical example - Zalman 7000Cu).

In describing each of the coolers I'll be drawing your attention to the fastening convenience. Also don't forget about such factors like fastening reliability, radiator size, material used, exterior, processing quality etc. Most of these are fairly subjective as well (especially, the exterior :), so I'll try to present as more illustrations as possible.

Cooler Master Aero7

By its exterior, the cooler is quite out of the ordinary - on top of its standard radiator there is a massive plastic structure with a turbine.

The turbine itself is just the distinguishing feature of the Cooler Master Aero7.

Its use allows to eliminate the so-called "dead spot", i.e. an area of weakest air circulation. Practically all the traditional fanned coolers some way or another suffer from this shortcoming. Some manufacturers are even doing their best to overcome this phenomenon (e.g. - Igloo 2320(Pro)), however, only Cooler Master was able to eliminate the "dead spot" issue completely.

To make this statement weightier, I performed the following experiment: installed a regular 70 mm fan (taken from the Igloo 2500 Pro) on top of the radiator of Aero7 cooler.

a hybrid of Aero7's radiator with an Igloo 2500 Pro's fan

After that, this hybrid ran for several hours in pretty dusty premises. Then I removed the fan and determined the airflow directions by the dust layer.

As is well seen on the photo, unlike the outer areas the fins on the central area of radiator have no dust layer. This means the air flow in the center is practically = 0, which certainly deteriorates performance.

Now taking a closer look at the turbine:

the range of rotational speeds varies between 1900 rpm and 4500 rpm, and the air flow versus rotational speed varies between 10.9 and 26.9 CFM.

The turbine overall dimensions are 70x70x70 mm. As a result, the cooler turns out to be too high and won't fit into some housings (those where the PSU is above the processor socket).

The turbine's rotational speed is set by the user. For that, there is a control unit (variable resistance) with a metal handle. We have come across something similar already (in particular, in Zalman and Thermaltake coolers), but the Cooler Master engineers brought the control convenience to perfection. First, the metal handle (also a plus - a cheap plastic part may have spoilt the exterior) is quite large, so it's simply a pleasure rotating it. Secondly, the speed control unit isn't loose deep inside the housing (like Zalman Fanmate) but can be installed either on the rear panel of the case (instead of an expansion card) or in the 3" FDD compartment. For that, there are two caps:

Keep it mind that the 3" cap is made of aluminum, so owners of housings made of this noble metal (among computer enthusiasts :) won't spoil the exterior of the front panel.

There is a regular molex connector used to power the cooler; besides, there is a special wire which can be plugged in to the motherboard and thus determine the turbine's rotational speed.

The radiator is fully made of copper, with the radiator base polished well enough (around 7/10 score points). Also of note is that all the fins have a small bend for better efficiency.

Finally, a few words about the fastening - the cooler is quite heavy (over 600 g), so it's no wonder the clip has a hook over all the three socket teeth (on each side). The fastening clip is balanced ideally in rigidity and elasticity. This makes the cooler installation and removal a breeze. A big part in that is played by the very long and wide finger stop - but it's just this part that might bring the Aero7 owners a lot of trouble.

First, on some motherboards this plate may hinder installation of memory modules:

Soltek KT400A motherboard

On motherboards having different socket positioning, other problems may come up. For instance, on the Epox 8RDA3+ board the socket is positioned too close to the edge, so the plate rests against the PSU (for the Inwin housing). But in larger size housings this is no longer a problem.

Chieftec housing

Now regarding the noise levels. In fact, you can't call Aero7 an absolutely noiseless cooler: even at the minimum rotation speed of the turbine you can hear noise variation during the short power-off (for about 15-20 sec). Then, as the speed goes up the noise is rising quite fast, and already at 3500 rpm it gets on your nerves. At the maximum rotation speed, the cooler noise fills your apartment all around:)

Thermaltake Volcano 11 Xaser Edition

Here is a flagship model among coolers made by Thermaltake for SocketA processors. I must admit I've been looking forward to the release of this model quite eagerly, since the combination of full copper radiator and a 80 mm fan gave every reason to expect a high efficiency. So once Volcano 11 appeared in the retail, I immediately ordered it.

The cooler is packed into a quite large box having two transparent windows: through the first you can see the cooler itself, and the speed regulator through the second. Like Aero7, this cooler allows the user to select the fan's rotational speeds.

However, the regulator in Volcano 11 is positioned on the rear panel of the case; and unfortunately there is no 3" panel. Actually, it's not hard to make it on your own - you've got to drill a hole of the right diameter and glue a radiator (which already has a glue base). Nice to note that the rotating handle is made of metal is large enough.

The cooler exterior is impressive - a copper radiator, black case (also a guideline for the airflow) and a formidable grating (that hides the fan).

The grating is in fact a decorative item rather than a real protection of fan against foreign matter. If you remove the grating, you will see the already familiar 80 mm Thermaltake fan of rotation speeds between 1300RPM and 4800RPM; the air flow varies between 20.55 CFM and 75.7CFM. But the noise levels vary between 17 dBa and 48 dBa at the maximum rotation speed. Frankly, I took the so high specifications, especially the maximum airflow (a bit too overstated, in my view) and the noise level (definitely understated) with some skepticism.

Then, if you remove the fan, you see a copper radiator with a metal frame attached on top of it. This frame guides the airflow and at the same time is a base for the 80 mm fan.

Removing the frame and taking a close look at the radiator. It has small dimensions (almost the same as in Volcano 7+).

The processing quality is very good, albeit not of mirror gleam. At the same time, grinding traces are visible on it.

Then, I didn't like the way the fins are fastened to the base. While in Volcano 7+ the radiator was something integral, the fins in this case are positioned mechanically on thermal glue or other thermal interface.

Once I detected this shortcoming, my joy of purchasing it almost vanished. It then disappeared completely once I started installing the cooler. The major issues are very tight clip (with a hook for 3 teeth) which is very difficult to catch on. Besides, the installation may leave scratches on the black cooler case, so it may lose its marketable look. Finally, to complete the installation, the user might need a screwdriver. At the same time, the clip has a stop with limiters (at least, this part of installation has nothing to complain about).

Once I started up the cooler I was really annoyed by the high noise level. While at the maximum speed the main noise component was air whistle, at the minimum speed the motor was rattling quite annoyingly.

By and large, this is not a big deal - we can replace the fan with another one (fortunately, there are lots of 80 mm fans out there on the market).

But replacing it with another one, we lose quite an interesting feature: the Thermaltake fan is able varying its speed depending on the processor temperature. For that, just plug in a special flexible thermal sensor that comes in the package bundle.

Along with the thermal sensor, the box has a tube of thermal paste, a Molex power adapter, and a brief description of the cooler and its operation modes.

Wrapping it up, I'd like to note that high retail price of the cooler which by the time it was tested amounted to $24.

Igloo coolers

In our tests, we used three Igloo coolers made by Glacialtech. These are Silent Breeze, Igloo 2500 and Igloo 2500 Pro .

The most powerful is the Igloo 2500 Pro model which we'll be reviewing in maximum detail.

Unlike the already reviewed Aero 7 and Thermaltake Volcano 11 coolers, this Igloo cooler is quite mean in exterior. But let's better start the roundup with the package box - it is really colorful and fully localized into the Russian language.

The radiator is rather massive, but let me put it straight that despite the uniform color, it is made of two materials: a copper plate and an aluminum base with fins.

Interestingly, the fins are of variable height,

and on top of it there is a metal plate to which a 70 mm fan is attached.

More details about the fan: its rotation speed is 4500 rpm at 35 dBa noise level. The airflow generated in the process amounts to 37.1 CFM.

Installing the cooler is less convenient: the fastening clip has a hook on the 2 socket teeth (which is quite enough - the cooler mass is merely 282 g), as well as a small stop/plate for fingers. The dimensions of the plate do not hinder installation of memory modules (the way it is with Aero7 cooler). But you will definitely need a screwdriver when removing the cooler - you'd have to hook the clip with a blade.

The processing quality of the base is good enough (7/10 score points); there is a layer of thermal interface on top. Remarkably, Glacialtech has given up using thermal interface with silver additives and switched to using more traditional thermal interface. To protect this layer, a plastic cap is put on the base.

Besides, on the base there are 2 strips (most likely made of rubber) which do not let the cooler get warped during the installation. Therefore, the probability of damaging the processor core during the cooler installation by a rookie user goes down.

As regards the Igloo 2500, by its exterior it is a mirror copy of the Pro model. Among the changes made is the yellow color of the metal plate and the less speedy fan running as fast as 2800 rpm. Of course, as the speed goes down the noise will also get quieter (down to 26 dBa), so will the airflow (down to 22.7 CFM).

Finally, here is Igloo Silent Breeze - again, it's an exact copy of the two previous coolers. What has been changed is only the color (the radiator has turned fully golden) and the fan parameters. Its rotation speed is merely 2000 rpm, with the generated airflow being 16.1 CFM. Clearly, with these characteristics the cooling is enough only for nominal operation modes, and overclocking through Vcore raise is out of the question. But in the case of Silent Breeze the developers posed a different task - to maximize the noise level. I must admit they managed to do a good job of that - as per the manufacturer's, the noise is below 19 dBa.

I can prove it through real-life tests: the noise from the cooler is completely missing (this is verified through the short-term power-off of the cooler for 15-20 sec). As regards the noise made by the Igloo 2500, even if it is there, it doesn't make much nuisance. But the noise coming from the 2500 Pro model is a bit discomforting.

Note that I test all coolers through using the Tt Speed Control speed regulator.

So, Igloo 2500 Pro run in the High-speed mode took a lead over Igloo 2500 and Silent Breeze by 4 and 8 degrees, respectively. Upon switching to the Medium mode the fan on the Pro model was running at ~3500 rpm, and the efficiency completely coincided with that for the 2500 model. The same was observed upon switching to the Low model: the Pro fan was running at ~2500 rpm, but the efficiency and noise fully coincided with similar parameters of the Silent Breeze model.

This is no wonder though because as I already mentioned, the differences between Igloo coolers are only in the color and the fan's rotational speed. Therefore, I do recommend to buy a cooler of 2500 Pro model only with a speed regulator. In this case, the user has more freedom in selecting the operation mode between the most powerful (albeit noisy) and completely noisy.

Thermaltake Volcano 9 CoolMod

The roundup of this cooler will be minimum: in fact, this is a modification of Volcano 9 which we have already tested.

All the specifications of both coolers fully coincide, with the only difference of the CoolMod model is in 4 additional LEDs installed on a plastic pad. Two LEDs (blue) are plugged to Power LED connectors positioned on the motherboard. The other two LEDs (red) are plugged to HDD LED connectors positioned also there.

As a result, once the system starts up, the interior of the PC is illuminated with smooth blue light with interim mixes of red (every time the HDD is addressed).

Basically, the idea is good, but the implementation leaves much to be desired. First, the LEDs aren't bright enough (especially, red ones), secondly, a couple of wires runs from each of the LEDs thus making up as many as 8 wires altogether. They are all loose, which makes a mess inside the PC housing.

On the cooler, there is a Thermaltake 80 mm fan whose rotation speeds vary between 1300 rpm and 4800 rpm. The air flow amounts to 20.5 CFM and 75.7 CFM, with the noise levels being 17 dBa and 48 dBa, respectively.

The speed is changed by either the user or automatically depending on the processor temperatures. For more details of the operation modes (and other specifications of the cooler), read the roundup of Volcano 9.

And a couple more comments: at the minimum rotation speed, the fan was running quite noisily (the noise coming from the motor). I do remember it quite well that while testing Volcano 9 the cooler was running almost noiselessly at the minimum speed (9/10 score points). It's most likely the quality of Thermaltake produce is getting worse. Along with that, the price difference between V9 and V9 Coolmod is around $4 (or ~$16 versus $20), which in my view is a bit too much for a piece of plastic with 4 LEDs on top.

Zalman CNPS6000 Cu

The Korean company Zalman can by right be regarded as a leader at battle against noise. All of its solutions are somehow or other targeted at reduction (or elimination) of this factor. However, to stay at the leadership the company has to work continuously at improving the produce. The company has done a good job of it for Intel Pentium4 processors (see the roundup of Zalman 7000Cu), but for AMD processors there hasn't been any novelties released so far. The best what Zalman can offer is the 6000Cu model.

The cooler sells in a smart box with a transparent window. Here are the contents:

• Radiator
• Fan of the fastening frame
• Pack of screws
• User's manual
• Fastening clip
• A tub of proprietary thermopaste
• Installation fixture
• Speed regulator: FanMate

The box may also contain a pack of silica gel so as to prevent the transparent window from misting during shipment through hot-weather countries :)

First, look at the radiator: it is heavy (462 g), made fully of copper (only the joining plates made of aluminum) and offers a good quality of the base.

All in all, as per AMD's specifications, the maximum admissible mass of the cooler should not exceed 300 g. Therefore, it's not recommended to transport the CPU unit with the radiator fit in place. Regarding that, there is an explicit warning for the users at the manufacturer's website saying that Zalman shall not be responsible for any damage to the system which may arise from inadequate transportation.

I said the "good" quality on purpose because first Zalman models (Zalman CNPS3000 Plus) used to have "superb" quality, almost mirror-like. But in our specimen it's simply smooth with grinding traces.

The base is made of plate edges joined with pins underneath, with the upper part branching like a sheaf (or a flower - Flower Heatsink). Therefore, a very large surface area is reached, which amounts to ~2900 sq. cm.

However, the design like this is not perfect, although it allows reaching greater surface area. The major shortcoming is this - the closer we move to the radiator base, the narrower the gaps between the fins are. As a result, this produces a "dead spot" in this place where air circulation is minimum.

In fact, a good radiator is still half the success - good-quality cooling is necessary. Zalman has also done a great job of that: the package bundle has a 92 mm fan whose rotation speeds vary between 1300 rpm and 2500 rpm. The airflow fluctuates between 31 CFM and 38 CFM, with the noise being between 20 dBa and 38 dBa.

I specify the ranges on purpose - the thing is, the package bundle has a device of speed regulation - FanMate. We have covered it repeatedly: a small plastic box with variable resistance. This regulator allows varying the fan's rotational speed very smoothly.

The FanMate also has a couple of shortcomings: the tiny handle makes the regulation uncomfortable. Besides, every time adjustment is needed you have to uncover the CPU housing since the FanMate is right close to the radiator. From this viewpoint, solutions made by Thermaltake and Cooler Master are more attractive because they allow bringing the cooler speed control off the housing boundaries.

The fan is mounted on a metal plate which is fastened straight to the PC housing. I have already mentioned the advantages of this cooling method - it's possible to fit additional fans on this plate in order to cool the other system components (e.g., it's the way I cool the video card and the system memory).

On the plate, there are long cuts for fastening the fan. The thing is because of different positioning of the processor socket on various motherboards, you have to adjust the fan's position. This can be done very fast since the fan is fastened by a big screw with special incision for fingers.

Thus, owners of the Zalman 6000Cu cooler can safely wait until Athlon 64 is released, because whatever processor AMD creates for the SocketA platform, its efficiency will be quite enough.

Tests

Before testing the coolers I summed up all their specifications into a table (the higher the score, the better).

Now regarding the measurement procedure. Without any noise measurement instrument on hands, I am going to assess the noise levels and rate it in score points. That is, the higher the score, the higher the score, the quieter a cooler is. Such assessment is quite subjective. I must the author (myself :) is very sensitive to the noise level since the computer is running on a 24/7 basis and must not distract from good night's sleep.

We performed the testing of coolers on the Soltek 75FRN2-L motherboard; one of the few boards based on the nForce II chipset that display the core temperature to the user. But to get that you would need the latest version of the Soltek HWmonitor utility.

Then, the testing system was assembled in an Inwin S506 housing, with the PSU brought outside. This PSU has a fan of variable rotation speed (depending on the temperature), and its close proximity to the cooler might affect the measurement results.

If a cooler had a related thermal interface, then the thermal interface was removed and further measurements were performed with the silicon-organic paste KPT-8 applied. The processor was heated up with the BurnK7 software, with a Barton processor used as the test processor. The operating clock speed was set to 2000 MHz (10x200) at the nominal voltage. Let's see how these coolers will cope with such a load.

Therefore, Cooler Master Aero7, Thermaltake Volcano 11 and Tt Volcano 7+ proved to be the most effective coolers. At the maximum speed, the core temperature does not exceed 60 C. But such high results were achieved by a high price: the operating noise exceeds all the reasonable limits. Of course the user will reduce the operation speed (all the three coolers are equipped with a speed regulator) to the medium or low level. As a result, the core temperature goes up, while in Volcano 11 it's rising sharply. So, we have two leaders only: Aero7 and Volcano 7+ of which the former appealed to me most.

By the way, CoolerMaster released an improved model of the cooler, Aero7+. Among the changes are the increased dimensions of the turbine - from 70 to 80 mm. In my view, this will allow reducing the noise level at the same cooling efficiency.

Then, as regards Volcano 11, I didn't like the cooler at all: expensive, noisy even at the minimum speed, poor design of the radiator. There's no other conclusion but this: don't buy it; the Volcano 7+ model looks much better.

As for the Volcano 9 Coolmod, - it depends on personal taste (e.g. owners of housings with a transparent window :). By its efficiency, it is equivalent to Volcano 9 (priced at $16) which in turn is equivalent to Volcano 7 (~$12.5).

Now on to the Zalman 6000Cu model - it demonstrated quite decent results both at the cooling efficiency and at the pretty low noise levels.

As regards the Igloo series, they are a good combination of moderate price, good cooling efficiency and low noise level. The latter factor deserves a special praise - Glacialtech engineers have done a good job making a high-quality 70 mm fan which is among the best at its performance. But use of the aluminum-copper radiator does not allow to attribute the Igloo cooler to the high-end rank.

Most likely, Igloo coolers fall within the mid-end models (or even value models; I have already said about the conditional nature of their division into categories), both at the price and the cooling efficiency. Igloo Silent Breeze which runs without absolutely any noise (noise levels 19 dBa versus 20 dBa in Zalman coolers) merits a special mention. However, I would advise to buy not this model but the Igloo 2500 Pro with some fan speed regulator.

Conclusion:

CoolerMaster Aero7

Pros:

• Excellent cooling efficiency
• Very easy to install
• Offers variable rotation speed
• Caps for the front and rear panels of the housing.

Cons:

• Problems of compatibility to certain types of housings
• High noise levels at middle and high speeds.

Thermaltake Volcano 11

Pros:

• Offers variable rotation speed

Cons:

• High price
• Very low efficiency at low speeds
• High noise levels
• Inefficient design of the radiator
• No 3" panel for the speed regulator
• Inconvenient installation

Zalman 6000Cu

Pros:

• Easy to install/remove
• Offers variable rotation speed
• Excellent noise/performance ratio

Cons:

• The speed regulator placed inside the housing
• Non-compliance to AMD specifications (primarily, by the mass)
• Problems of compatibility to certain types of housings

Glacialtech Igloo 2500 Pro

Pros:

• Good noise/performance ratio
• Relatively low price
• Convenient installation

Cons:

• Does not meet the high-end level (i.e. the overclocking margin is restricted).

So, the totals have been summed up. Now turn to a very interesting data on the difference in temperature readings taken from the external thermal sensor as opposed to the internal. The amount of the difference is not constant and strongly depends on the cooler used.

The higher the factor, the more difficult it is to give an adequate estimation of the cooling efficiency on those motherboards which can't display core temperature data to the user.

Note that practically all motherboards built on the nForce II 400 Ultra chipset feature hardware protection against overheating, but only few are able showing the real temperature to the user. It's not right to use the data read from the external thermal sensor for hardcore overclocking, since they don't show the factual processor core temperatures.

Read more on this topic:

Address your questions, wishes and suggestions to the conference.

The author would like to express his appreciation to Sunrise company for Igloo sample coolers presented for tests.