Pentium4 Coolers Roundup. Q4 `2002

Introduction

Coolers Roundup: Pentium4. Q3 `2002 SocketA. Q4 `2002 SocketA. Q2 `2002 Testing thermopastes

In the previous review of new coolers for Socket478 we examined some interesting models. And we made a quite important conclusion: if the system doesn't overclock or if overclocks moderately (with a slight voltage raise), a box-type cooler will do. It runs quietly enough (since the fan's rotational speed is variable depending on the temperature) and cools well. But if you are into serious overclocking, then you won't get by without a powerful cooler and a fully (or partly) copper radiator. For that, Thermaltake Volcano 7+ has shown itself as a good performer.

Today we'll be looking into more coolers with the aluminum radiator and the same amount of copper-based coolers, among them "the heavy gun" - Zalman 6500B-CU.

A couple of words on the measurement procedure. Like in the previous review, I will estimate most of the subjective parameters with score points. The higher the score, the better a certain parameter is. The subjective parameters are first the cooler's noise levels, ease of installation and further removal. Besides, we'll measure the processing quality of the base also in score points.

But the weight, prices and temperatures are precise enough and will be given in figures. Well, the processor temperature and the fan rotational speed were measured with the Motherboard Monitor v5.2.1.0. The test setup was assembled in an Inwin S506 case with the cover removed (for better perception of the noise :). If the cooler had a related thermal interface, then the latter was removed and further measurements were carried out with the silicon organic paste KPT-8 (as the cheapest among all the tested pastes). The processor was heated up using the CPUBurn software, with Pentium4 1.8A used as the tested processor (all Pentium4 processors have thermal sensors integrated into the core). For increased heat emission we set its FSB speed to 133 MHz (which totally gave 2.4 GHz) and raised the Vcore voltage up to 1.65V. To explore the behavior of coolers during extreme overclocking, I set the FSB to 150 MHz and raised the voltage to 1.85V.
The overclocking was performed on the Asus P4GE motherboard (based on the Intel 845GE chipset).

While describing the cooler I'll draw your attention to the fastening convenience and reliability. Also don't forget about such factors as the radiator's dimensions, material used, appearance, processing quality etc. Most of these are fairly subjective (especially, the appearance :), so I'll try to be as impartial as possible through presenting more illustrations.

Arctic Storm 2

I first thought the cooler was a noname product, but it turned out to have a manufacturing company (although no mention of that was on the package). Arctic's website is available over here. Besides the cooler, the box contained metal latches for the socket and a tub of thermal paste.

As I already said, the cooler looks quite impressive (its dimensions are 83x69x55 mm).

First, the impression is produced by the radiator shape with a couple of dozen fins placed on a pretty thick base.

By the way, the base is of varying thickness - it's thicker in the central part.

The lateral walls of the radiator have deep vertical and fine horizontal saw cuts - every inch of the surface is used for increasing the air contact area.

The processing quality of the base is middling.

On the radiator there is a black plastic frame on which a 70x70x15 mm fan is fastened with four screws.

By its noise characteristics, the cooler is pretty inferior to its box-type brothers - its noise is quite a nuisance. With the rotational speed 4500 rpm declared by the manufacturer, its noise level is 32 dBa and the air flow is as much as 25.73 CFM
In practice, the fan was rotating at the speed 4000 rpm.

The cooler fastening system left a dual impression. It is made up of two standalone metal clips allowing fastening the radiator in the socket (using the hooks in the plastic frame). Such a scheme is appealing because of its simplicity - it takes merely a few seconds to attach and remove the cooler. But the clamping effort produced is not much great, and the look-and-feel does not bring much confidence.

Finally, the price for the cooler is approximately the same as for a "box-type" cooler and is around $10 a piece.

Igloo 4300 & Igloo 4200

We are testing the two models: Igloo 4200 and Igloo 4300. The first is pretty quiet, the second is a bit noisier but offers higher performance.

All in all, the only difference between the two coolers is their fans (as well as their layout).

The first model (Igloo 4200) features a rather unusual placement of the fan - it is shifted by some distance from the transverse axis of the cooler. Such asymmetrical position allows minimizing the "dead spot" and improving the fan's efficiency.
This is very important for any cooler, especially for the Igloo 4200 since there is a low-speed fan on it (the speed declared by the manufacturer is 3000RPM, and the dimensions 60x60x20 mm). The low speed entails poor cooling efficiency (the air flow is 13.3CFM), but on the other hand this suggests a low level of noise (merely 25dBa).

As for the noise, the Igloo 4200 cooler did appeal to me - it runs a bit quieter than the "box-type" brother. But the fan of the Igloo 4300 is much noisier than its younger brother. The declared rotational speed is 4800 RPM, and the air flow is already 30 CFM, with the noise level 37 dBa.

But the much increased air flow is not caused by the increased speed - the fan's dimensions in Igloo 4300 are a bit greater - 70x70x15 mm.

And finally this fan is placed precisely at the center of the radiator (unlike Igloo 4200).

Here is a photo vividly demonstrating the differences between the Igloo 4200 and Igloo 4300 coolers.

The remaining components are practically the same - the metal frame on which a fan is fastened, has a handy mechanism for attaching the cooler to the socket.

This mechanism is a latch whose single motion is enough to fix the cooler firmly in the socket. Removing the cooler is a matter of a few seconds.

As you can see, the latch with the fan is not connected to the radiator - it is removable with a single motion.

By the way, on this photo you can easily see a transparent plastic cap on the base. It is meant for preserving integrity of the thermal interface layer which contains silver as GlacialTech declares. I have no idea how much silver is there, but it looks like that by appearance. In any case, by its characteristics this thermal interface is no better than the domestically produced thermal paste KPT-8.

As far as the radiator base is concerned, the processing quality is better than in the cooler by Arctic. But it can't be regarded as ideally smooth - grinding traces are anyway visible.

The radiator itself has around 30 fins of varying height, with each fin being trapezoidal.

Prices for these coolers are approximately equal to those for the "box-type" cooler and are as follows - ~9$ for the Igloo 4200 model, and ~10$ for the Igloo 4300 model.

Igloo Diamond 4000

The advantages over those mentioned are visible straight from the package already. The package is abundant in colored pictures, tables and diagrams praising the Diamond 4000 cooler (even the letters are in raised bold).

On the reverse side of the package there's a template saying the maximum processor clock speed the cooler is suitable for is as high as 3GHz.

Not quite clear if the cooler can be used with the most recent 3.06 GHz processor (whose clock speed is higher than 3 GHz :). Well, we've already seen that the real Igloo coolers specifications did not meet those declared.

Another interesting item in the table is the phrase mentioning of the "five-year guarantee" for the cooler. Strangely enough, how to get the guarantee obligations implemented if the retail shop gives a 2 week guarantee at best.

I wonder if such a drastic reduction in guarantee periods is only a domestic computer sales specifics? Or they do it that way all over the world? :)))

And finally the most interesting piece of information is the line "Rotational speed 2800-4800RPM". Wow, that means a fan of variable rotational speed!

Enough with examining the package though - let's take out the cooler.

Regretfully, the photo can reproduce this effect - all cooler parts are gleaming but for the fan. And the fastening plate is polished to the mirror-like gleam.
The only thing I disliked about it was the color spectrum. Pure bright colors would look much better (a matter of personal taste though). Forget the exterior though. What matters is the cooler does its job well - cools well.

Let's take a look at how the function for speed variation depending on the temperature is implemented.

For that, a thermal sensor is used (shown in green color on the photo). Draw attention to its layout - it is placed right on top of the fan and has no connection with the radiator.

That means, it varies the air temperature, and considering the air flow direction the thermal sensor changes the air temperature inside the PC case.

Among the numerous tested coolers we've already come across a similar thermal sensor. That was ThermalTake Volcano 7 whose thermal sensor also changed the air temperature inside the PC case. And it was calibrated in such a way that poor users had to devise a means to cool the thermal sensor in order to reduce the rotational speed (and more important, the noise) of the master fan.

The Igloo Diamond4000 cooler has the reverse problem. You don't have to bother about the quietness - at its minimum rotational speed (2800 rpm) the fan runs very quietly (23 dBa), and the air flow amounts to 12.4CFM. At the maximum rotation speed (4800RPM), the fan should provide the air flow at least 22.8 CFM with the noise level no more than 35 dBa.
The problem is different - during the tests the fan wouldn't run at its maximum rotational speed, even though the processor had been overclocked through voltage raise and, but it seems like the processor's thermal protection against overheat worked. Just the very same time the Diamond4000 cooler regarded the processor's no-rising temperature as its own achievement so there wasn't any need to increase the rotational speed.

If the thermal sensor had measured the radiator temperature, the rotational speed would have been rising in direct dependence on the processor load. It turns out the fan's rotational speed depends on the ambient air temperature, and not the processor temperature.

As for the base, its processing quality is excellent. No grinding traces were found. Like in other coolers made by Glacialtech, on the base there is a layer of thermal interface protected with a plastic cap.

The radiator design fully coincides with that for radiators in the Igloo 4200 and Igloo 4300 coolers.

Here is the conclusion. Diamond4000 is an Igloo 4200 cooler with the better processed surface and having a fan that varies the rotational speed depending on the air temperature inside the PC case.

Wrapping up with it, the price for this cooler is around ~$14, which is too much for a cooler of this class. As we see further, with this money you can buy a cooler having a copper radiator.

Titan W5TB(TC)

Today we'll be looking into one more cooler, also on the aluminum radiator, but featuring some interesting peculiarities.

Let's start with the package bundle - the box is of familiar look-and-feel, besides the cooler itself there is a syringe of proprietary thermal paste, fastening clips and an installation instructions slip.

In fact, the slip was of no use - the cooler installation can't be simpler. The fastening is very similar to that used in the Arctic cooler - the fastening clips are standalone. However, in our case the clips have additional stiffening fins and fix the cooler firmly in the socket. There is big finger stop on the clip. This makes the cooler installation and removal a matter of a few seconds.

The very first powering of the cooler pleased me with completely noiseless operation. The noise was coming from all around: the video card cooler, the IBM hard disk, but not the processor cooler.

Just for curiosity, I disabled the hard disk drive, replaced the video card with a Elsa Geforce2 MX (in which there is a passive radiator). But anyway the noise stayed - the noise was coming from the power supply unit, but the Titan W5TB(TC) was still noiseless.

Thanks to the outstanding noise characteristics, of most interest is the fan, of course.

Its dimensions are these: 70x70x25 mm, the air flow between 20 and 33 CFM with the noise levels not greater than 22-28 dBa. As you have already guessed, the fan features variable rotational speed varying between 1900 and 3300 rpm.

The rotational speed depends on the temperature measured by the thermal sensor. Unlike the Igloo Diamond4000 cooler, the thermal sensor is placed on an extender running deep into the radiator through a special slit. And the thermal sensor is attached directly to the radiator base (for that, thermal glue is used).

the thermal sensor placed on an extender

the slit for the thermal sensor

Such a fastening method provides a more precise dependence of the rotational speed on the processor temperature (not the air temperature, as in the Diamond4000).
By the way, the Titan's website gives a table with approximate description of that dependence. And the last note regarding the fan - if the user decides to plugging it in via a speed regulator (sort of Tt Fan Speed or Zalman FanMate), that should be done very carefully. The thing is the 5-7V voltage is evidently not enough to the fan and it halts.

The fan is fastened to the base with a special metal frame (which is also a stop for the fastening clips) and four screws.

Regarding the radiator itself. it appears to be a pretty thick base on top of which there are 27 absolutely identical fins installed.

Draw a special attention to the processing quality of the base - it is very smooth, almost of mirror-like gleam.

What is interesting, the lateral walls of the radiator are also polished to the mirror-like gleam. People at Titan must have found a very cheap method of quality grinding and are using it in all parts whatever falls into hands :))

Certainly that does not bring any practical advantage from that, but the exterior certainly wins from that! That seems to be the nicest cooler for Pentium4 processors. A couple of fine details to note. The fan has a silvery coating (quite easy to scratch - so be careful!) and a protective wire grid.

Wrapping up with it, note that the price for this cooler is around $10-11.

Besides, recently an enhanced model of the Titan W5TB(TC) cooler has hit the retail shelves. First, its thermal sensor is placed in a special opening in the radiator base (which should improve the processor response to the temperature variations).
Secondly, the cooler sells in a much smarter box with a transparent window.

Titan CW7TB

But let's start with the package bundle - the cooler comes bundled in a regular cardboard box together with a proprietary tub of thermopaste.

Like GlacialTech's coolers, the Titan CW7TB model is made up of two independent components: a radiator and a fastening frame with a fan:

It's praising to note how the cooler is safeguarded - the radiator is shielded with foam-rubber all around (but for the top).

Moreover, under the fan there is a special foam-rubber washer (should be removed prior to removing the cooler).

People at Titan must have seen of the way our dockers work :)))

Well, taking out the radiator. We are facing the second protection frontier of the radiator - the polyethylene film.

Removing it, and - there we are with the radiator in all its glory.

It's now clear why it has been looked after so cautiously. The surface of the base is ideally smooth, of mirror-like gleam. The lateral fins are made at the equal quality - you can even see the reflection of yours in it.

To prevent the fins from bending and sticking, they are separated by small pressed-out projections (from above they look like copper tubes).

Now taking a closer look at the fan. It is made multifinned, of small bulk (70x70x15 mm) and runs pretty quietly. As per the manufacturer, its noise levels are below 28 dBa at the rotational speed 3500 rpm. And it provides air flow that amounts to 27.96CFM.

The fan is fitted on a rather massive plastic frame that has a pretty tricky mechanism for fastening to the socket (BTW: the mechanism is patented).

Fitting the cooler is a breeze. The only thing to keep in mind is that it should be fitted with one side first with the other latched afterwards (trying to latch both sides at the same time fails).

The exterior of the cooler is quite impressive. A special trait is the protective grid fitted on the fan like on other Titan-made coolers.

Resume: a quite probable candidate for victory by its "price/performance" ratio. The price of the cooler is about ~ $12.5 (cheaper than for Diamond4000!).

Zalman CNPS5700D-Cu

But - all in good time. Let's start with packaging.

The cooler sells in a transparent package to allow the buyer to have a better idea of the design prior to purchasing. Once you hold the box in your hands, you immediately feel how light and small it is. Just fancy that! - the cooler weighs about 500g, which exceeds the 450g limit stipulated by Intel specifications. By the way, both on the website and in the documentation there is a special address to the user to take precautions when carrying the computer around (besides, Zalman shall not be responsible for any damage if the cooler happens to fall off :).

As the sticker says, the cooler is useable for processors having clock speeds up to 2.8 GHz and higher.

Besides the cooler itself, the package contains a tub of thermopaste and a device for controlling the fan's rotational speed, Zalman Fanmate. The package also contains the cooler description in English and Chinese (maybe it's not Chinese, but written in hieroglyphs :).

Well, taking out the cooler - in the daylight it looks simply fantastic.

To have a better idea of its design, we're dismantling it down to the last part.

Like all the previous Zalman-made coolers, the radiator is made up of numerous copper fins (I even wouldn't count them all :) tightly bundled in the center. The structure is fixed with a metal cramp having stiffening ribs. The net effective area of the radiator is enormous, which amounts to 1270 sq.cm.

A couple of things to note: first, the radiator base does not completely cover the processor. This is not a big deal, since the processor core is much smaller than the IHS (Integrated Heat Spreader, a copper plate on the processor), but the hardcore overclocker will anyway regret not having those small sq. mm's that are not part of the heat transfer process.

Secondly, the base is polished well, but not to the perfection.

The copper radiator is enclosed in a plastic case on top of which a fan is fastened in its turn. For that, 4 long screws are used.

The fan is an ordinary 80 mm ZM-F1 thing made by Zalman (which can be purchased separately). Like the 92 mm (ZM-F2) fan, it has proved to offer reliable and quiet operation mode. The noise level is over 20 dBa at the rotation speed 1700 rpm. At the maximum rotational speed (3100 rpm) the noise level is higher and amounts to 34 dBa. The speed variation is performed manually, using the Fanmate device.

A special base made of transparent plastic is fastened with the same screws.

And on top of the base the airway is fitted.

The airway is fastened with three screws. Note that along the edges of the airway there are special slits which allow changing the rotational displacement. Indeed, considering the variety of PC cases, the function of airway rotational displacement is a definite necessity.

Well, the cooler has been assembled (reminding you that it sells pre-assembled, we simply amused our curiosity :). Now let's get down to fitting it on. Just a note: the fastening procedure is a bit unclear: while pressing it, thread the fixed teeth through, then holding the cooler pressed, snap the mobile teeth (or latches) neatly.

The cooler is easiest to fit on a removed motherboard. So I did that. Great! The cooler is fitted safely in place, but the problems still persist. It turned out while installing the motherboard with the cooler that the airway base was resting against the power supply unit and no way was there to avoid that.

Just for reference - the cooler dimensions are 125x125x150 mm.

My case is a standard ATX (Inwin S506). The socket placement on the motherboard (Asus P4GE) is also pretty standard. This is why so many owners of ATX cases can't enjoy the benefits of operating the CNPS5700D-Cu cooler, regrettably.

In despair, I removed the airway and tried using the cooler in the traditional manner. But nothing good came of that - the cooler wasn't much better at its job than the box-type cooler. I even changed the air flow direction by putting the fan upside down. But that did not help improve the cooling.

It should be pointed out though that cooler was blowing not to the radiator, but in the reverse direction. The air thus heated is expelled from the case via the airway.

In gloomy mood I started thinking (which I should have done at the very beginning) and thought out something. The solution was simple (and ingenious :) - you've got to remove what hinders, i.e. the power supply unit. Of course, not all had to be removed, but it had to be brought outside the PC case.
That simple manipulation allowed fitting the cooler in place and testing it successfully. By the way, the advantage of the step is that much inner space inside the case was freed up. First, that facilitated removal and installation of the boards. Besides, I got an opportunity to remove and fit coolers having rather tricky fasteners (before that, I had to remove the motherboard from inside).

Here is what we've got:

This is an absolutely operative system for the expulsion of upheated air off the PC case. The additional fan is a compulsory item in the system; otherwise, the whole idea following which the CNPS5700D-Cu cooler is built breaks down.

If it seemed to somebody that the additional fan was shining, this is true :) Anticipating your question as to where one can buy such a cutie, here's the answer - at the computer market in Taipei.

And the last but not least, the noise levels. The noise itself coming from the 80 mm thing is not that great, and it is pretty quiet in the "slow" mode. But the airway installation had a most negative effect on that - the noise went up essentially. Installing an additional fan doesn't help reducing the noise, it simply can't be eliminated - that is a compulsory condition for any efficiently running system.

The retail price for the cooler is around $33, however, to get full-featured cooling you have to add the cost of additional fan (fitted on the rear case wall).

Zalman 6500B-Cu

Well, let's start examining the cooler from its packaging.

As the sticker on the box says, the cooler is useable for processors of clock speeds up to 2.4 GHz and higher. But no mention was regarding how "higher". However, as our tests will tell, the 6500B-Cu cooler has it all to cool even the fastest and most powerful processors.

The box contents are already familiar to you from the previous review on Zalman coolers. These are the cooler's description, a device for controlling the fan speed - Fanmate, a tub of thermopaste and a pack of screws to fasten the bracket with a fan.
Here is the bracket, or the "foot", "Zalman's foot" ( a more common nickname :)

On the "foot" there is the already familiar 92 mm fan rotating at 1600-2500 rpm (controlled by the Fanmate). The noise level is within 20 and 33 dBa, with the air flow being between 31 and 38 CFM.

Besides, in the box there is a radiator, fastening clips made of blue plastic and a couple of metal washers.

Now on to the "spice", - the radiator.

As I already said, the mass of the radiator (898 g) is staggering. In all the other respects, it is a usual structure - its numerous fins are fitted into the base with screws and then sheaved up (or as decided at Zalman, "flowered"; which gave the name for this structure - Flower HeatSink).

The dimensions match the weight. On the photo, the radiator is placed close to the 2-ruble coin (believe me, the coin is very small in size). For reference - the surface area of the radiator's copper part is 3400 sq.cm.

And when the radiator is fitted in the socket, part of its fins exceed the bounds of the plastic frame. In some cases, this may hinder memory modules installation.

Look at the radiator's fastening scheme - it is fitted in the socket with two plastic latches, which is easy to ascertain, but very difficult to remove (you have to press upon the radiator and the latch quite heavily). If the user happens to think of the clamp as being not strong or reliable enough, he can put metal washers underneath the latches.

As for the base, I did not like its processing quality: it is smooth, but the grinding traces are anyway visible. The base in the first Zalman models had almost mirror-like surface.

Finally, the system is assembled.

For those who first use the Zalman coolers - let me explain that the radiator and the fan are absolutely independent on one another. First you fit the radiator, and then fasten the fan on a bracket (or, "a foot"). The user can also vary the fan's position within quite wide ranges (depending on the position of the processor socket on various motherboards).

By the way, the "foot" is very handy for installing an additional fan to cool the video card (the photo below best illustrates this).

It's a pity I had no Celeron Northwood processor,- I did want to conduct a sort of experiment on whether it is possible to set up passive cooling of the processor using the radiator alone.

The price of the cooler is about $40, which is quite acceptable for an almost a kilo copper radiator.

Zalman 6500B-AlCu

The box contents is absolutely identical to that of the 6500B-Cu model: a user's guide, a Fanmate, thermopaste and the fastener.

Since all the differences from the "elder" brother are in the radiator, let's dwell on it in more detail.

As for the base processing quality, it's the same as in the Zalman 6500B-Cu model, the surface is smooth, but not mirror-like. In my view, the presence of copper fins is questionable - the copper plate in the base is too thin.

As we see, the number of copper fins (or petals) is cut down to 8. This resulted in mass reduction and better heat dissipation. But the price is a bit too high - $33, which is too expensive.

Tests

Notes: * The fan changes its rotational speeds with temperature. ** The noise is absolutely unbearable! The rotational speeds, air flows and noise levels are as per the manufacturer's data. It has passed quite a long time (in the computer industry scales :) ) since we issued our previous roundup of coolers. Several new processor models have been released, among them the Pentium4 3.06GHz HyperThreading processor. Yes, clock speeds do go up quite fast, and so does the heat emission. That's why Intel has plans to release a new box-type model of their cooler having the copper base. Therefore, many models that used to look quite competitive now can't cope with the job of high-quality cooling. The High, Medium and Low modes were set with the Tt Speed Controller switch The results are quite evident - coolers with copper radiators do their cooling job better than their brothers made of aluminum. A really remarkable thing is that the Arctic Strom2 cooler proved to be a much better performer as compared with the Titan CW7TB cooler that demonstrated absolutely poor results. The difference was so big that I did repeated tests which confirmed the original conclusion: thanks to its more powerful (and thus more noisier) fan, the aluminum Arctic Strom2 does a better cooling job than the copper Titan CW7TB. That's what using a weak fan on a good radiator may result in :) Just for experiment I removed the fan from the Tt Volcano 7+ cooler and fitted it on the Titan CW7TB. Here is what I've got: in the table and graphs this hybrid is dubbed as "Titan mod" This structure was as noisy as Volcano 7+ (which is natural), but it cooled by 1-2 degrees worse. It's possible that the radiator of the Tt Volcano 7+ does a better job cooling the processor owing to the much thicker base and a slight bend of the fins. The next stage during the test was cooling the processor overclocked to a higher extent. The FSB was set to 150 MHz and the processor voltage to 1.85V. As a result, the processor was running at 2.7 GHz. Only coolers having copper radiators were allowed into this test. And considering that the Pentium4 features a thermal protection mechanism, upon completion of the Burn software I started a program for picking the keys to a cipher (or something like that) and was watching the results. The most important thing in the program is that every second it displays information on the number of keys processed. In the nominal mode, the number of processed keys was approximately 1960+-30. In case of overheat (when I disabled the fan) the thermal protection mechanism came into play, and depending on the cooler installed the performance was gradually going down to 1300 keys/sec, after which I enabled the fan again. As a result, the performance was coming back on track. Even installation of the Zalman 6500B-Cu monster resulted in performance drop by 100-120 keys/sec. Introducing an additional fan (fitted on the same "foot") didn't help improve the results. And only the replacement of the 92 mm fan with a 120 mm thing allowed cooling the processor to the normal limits. The same results were demonstrated by Tt Volcano 7+ activated to the full sway. But in testing the Titan CW7TB cooler the temperature did not exceed 61 C thanks to the thermal protection; and the performance dropped down to the ~1500 keys/sec mark. Therefore, the conclusion is this - to raise the Vcore > 1.8V you've got to provide a very high-quality cooling. the 120 mm fan is a real monster :), to the left - there is a 92 mm fan, and on the rear wall - a 80 mm fan. The Zalman 5700D-Cu cooler made a very good showing - no performance drop was observed. And the cooler proved to be a much better performer in a closed case - it was able to outperform even the Zalman 6500B-Cu. The processor clock speed = 2.52 GHz, the Vcore = 1.725V, the case closed. So here are the brief findings. Overclocking and cooling the Pentium4 processor is not as easy as it may seem at first glance. There are lots of nuances to be considered. Findings If you intend using the processor in its nominal mode, you can safely use a box-type cooler. But if the noise levels are very critical to you, turn to coolers of varying rotational speeds. Some general recommendations for them: if the processor is weak (that could be a Celeron Northwood or lower-end Pentium4 Northwood models), take the Titan W5TB(TC) without any fear. This cooler is very(!) quiet. If the processor is bit more powerful (or if you plan to upgrade to one in future), you are better off taking the "box-type" cooler (which is also able varying the rotational speed depending on the temperature). I wouldn't recommend buying the Igloo Diamond4000 because of its poor job adequately cooling the processor. Besides, the cooler is pretty expensive for its category. The Zalman 6500B-AlCu model is more expensive than that, but it cools a little bit better than the box-type cooler. I wouldn't recommend to buy it either. If you plan for a slight overclock (i.e. the Vcore voltage isn't going to exceed the threshold 1.65V or +10%), then I can't give a one-one recommendation. As an option - if you have a good 70 mm fan, then take the Titan CW7TB model and replace the factory fan with your own. But if you are into overclocking with raising the processor voltage by more than 10% (i.e. Vcore>1.65V), then an expensive cooler with the copper radiator is a must-have. The cheapest in this category is the Thermaltake Volcano 7+ model which does best at the maximum rotation speed. But the noise it makes is simply unbearable. If we switch the speed to the medium mode, the noise will go down to the acceptable level, but its cooling job will be worse than the copper coolers made by Zalman. As for the Zalman 5700D-CU and 6500B-Cu coolers, you have to take account of the following factors when deciding which to buy: fitting the Zalman 5700D-CU inside big-tower cases is a breeze; the Zalman 5700D-CU requires an additional fan doing the job of blowing-out and placed as close as possible to the airway outlet; the Zalman 5700D-CU is noisier than the Zalman 6500B-CU; in the closed case, the Zalman 5700D-CU shows better results; The radiator of the Zalman 6500B-CU cooler is very heavy - twice as heavy than recommended. The Zalman 6500B-CU cooler is the most expensive; in the closed case, the Zalman 6500B-CU requires good ventilation; in the open case, the Zalman 6500B-CU provides the best cooling. As you can see, it's hard to make a one-one selection (but I would rather choose the Zalman 6500B-CU). If you are into the maximum overclock with a very quite operation of the cooling device, then forget about the air cooling. Addendum In the nearest time, some very interesting coolers are expected to hit the retail shelves. This is the Zalman CNPS7000-CU model, which by its design does resemble the notorious Orb's :) And a couple of novelties from Thermaltake, namely Spark7 and SubZero. What is more, the retail has been offering the Cooler Master IHC-L71-E1 for quite a while, made on the base of the "heatpipes", but so far it hasn't yet fallen into my hands :). As soon as the good chance appears, we'll test these. Read more on this topic: Coolers for Pentium 4 `2002 Q3 SocketA Coolers Roundup Q4 `2002 SocketA Coolers Roundup Q2 `2002 Testing thermopastes