by Sergey Lepilov
03/24/2009 | 07:06 PM
We have been introducing new CPU cooling solutions to our readers on a regular basis. And although we are proud of having tested almost everything that is available in the cooling solutions market, these reviews are not perfect for one simple reason. The thing is that we compare only 2-3 cooler within each review taking one of them as a reference for our performance analysis. That was the case when Thermalright Ultra-120 eXtreme rules the world, and also when ThermoLab BARAM took over the lead for a short while until Scythe Mugen 2 came out. With one same cooler participating in each test session, even with a small measuring error margin, it is possible to compare the efficiency of cooling solutions from different articles.
Nevertheless, time has come for a big comparative analysis of the coolers performance, because there are a lot of highly efficient solutions available in the market these days. The final factor that determined the need for this test session was the launch of the Intel Core i7 based platform with the new LGA1366 socket. Today we are going to discuss how the new and old super-coolers will perform in the new platform, what the peculiarities of Nehalem core cooler are and how the Intel Core i7 overclocking potential depends on the cooling efficiency.
We are going to open our today’s article with a detailed review of a new cooling solution - Prolimatech Magahalems – that is a new candidate for the super-cooler title.
The company called Prolimatech Inc. is not very well-known. It was founded in Taiwan only a year ago, but has dedicated professionals with 22 years of experience in the cooling field in it. Their first (and so far the only) cooler is called “Megahalems” which sounds similar to the new processor microarchitecture name – Nehalem. We don’t know if they chose this particular name on purpose, but, in fact, it is not really important. Just interesting.
The cooler ships in a small box with a convenient plastic carry handle:
There is a polyurethane foam casing inside the cardboard box that holds the heatsink and accessories sealed in small plastic bags. Among them are:
As you can see, there is no fan shipping with Megahalems, so you will have to purchase one separately. The cooler is made in Taiwan and its recommended retail price is $59.95.
Prolimatech Megahalems is tower-heatsink measuring 130 x 74 x 158.7 mm and weighing 790g. The heatsink sits on six copper nickel-plated heatpipes 6mm in diameter that go through a copper nickel-plated base.
The heatpipes hold two aluminum heatsink arrays with 5mm gap between them.
According to the manufacturer, the heatsink was designed of two separate arrays in order to reduce the resistance to the airflow from the fan and not to eliminate the “dead zones” in the heatsink, as you may have thought in the beginning. For the same exact reason the heatpipes inside the heatsink are all lined up. As a result, Prolimatech engineers can claim that their cooler is optimized for fans with 800-1200RPM rotation speed, although it is not quite so in reality. Moreover, having heatpipes lined up into a single line doesn’t ensure even heat distribution over the heatsink plates. As a rule, coolers with design like that yield to their competitors with more optimal heatpipes positioning.
Each heatsink array consists of 45 aluminum plates, each about 0.5mm thick:
The heatsink has a total of 90 plates, so its calculated effective cooling surface taking into account all the curves, sides and heatpipe holes makes 7008 sq.mm, which is a pretty average size for contemporary cooling systems.
If you take a closer look at the cooler, you will see that the heatpipes do not pierce the heatsink plates, but the plates each consist of two halves that are soldered to the heatpipes and clipped together on the sides:
Of course, this contact method does have an evident advantage over traditional one, when the plates are pressed against the heatpipes piercing them. Since the plate is cut in half, there is no need for technological holes, there is bigger contact area between the heatpipes and the plates, and hence the cooling efficiency increases. Besides, Megahalems has pretty thick heatsink plates, so it could be pretty hard to press them against the heatpipes tightly enough.
We also noticed that the manufacturer tried to bend the heatpipes as little as possible making sure that there is no additional resistance to the heat flow inside the pipes.
They resorted to soldering technique not only for the heatpipes and heatsink plates, but also for the heatpipes and the base of the cooler. You can clearly see solder traces on the sides of the base plate:
To ensure the most effective heat transfer, the heatpipes have been put into special grooves in the base plate. The thinnest part of the plate beneath them measures 3mm.
The base finish quality leaves much to be desired; there is no polishing of any kind. Moreover, you can feel the machine marks if you touch the base plate surface.
However, it is not as important as the evenness of the cooler base. Let’s check it out now:
It turned out that the base of the Prolimatech Megahalems cooler we got for review is slightly protuberant and one of the four corners is dropped by about 0.25mm. Despite these observations, the thermal compound imprint on the CPU heat-spreader turned out quite satisfactory:
Cooler installs in a very simple and intuitive way. First you insert the spindles through the backplate and mainboard PCB, then set two metal retention brackets on top of them. After that the cooler is placed on top of the CPU and is pressed with the third bracket with spring-screws:
That’s all. Very simple and reliable. And two rods on the pressing bracket that go into special holes in the upper part of the cooler base will prevent it from rotating on the CPU heat-spreader. Looks like Prolimatech Megahalems has the best retention solution of all coolers we have tested so far: it is extremely easy and convenient to install and presses the cooler against the CPU heat-spreader with just the right force. The only thing we could probably find inconvenient is the fact that you won’t be able to remove the CPU once the retention for Megahalems cooler is in place, because the locking lever will hit against one of the retention brackets when in the open position. However, these brackets are fastened with screwed caps without any special tools, so it will only take you 30 seconds to remove then and take out the brackets.
Prolimatech Megahalems installed onto a board will not interfere with any electronic components around the processor socket. The distance between the lower heatsink plate and the mainboard PCB surface measures 45mm:
You should use two enclosed wire clips to attach the fan to the heatsink. They go into the retention holes in the fan frame and catch on to the special groove in the heatsink array:
Since the heatsink is symmetrical and the grooves are available on both sides of it, you may install two fans at the same time. However, the additional two wire clips that you will need are not included with the cooler and need to be purchased separately. It is actually a pretty strange drawback for a relatively expensive and ambitious cooler like that.
When the cooler is installed onto the mainboard so that the airflow is directed towards the case back panel, the cooler fan blocks the first memory DIMM slot on the board and it is indeed the case not only for the memory modules equipped with tall heat-spreaders, but also with the memory modules with regular or no heat-spreaders at all.
If you install the cooler so that the airflow is directed towards the top of the case, then the first memory slot may be used for memory modules without heat-spreaders. By the way, the cooler is most effective when it is installed in this particular way than the way on the photo above:
It is pretty important to achieve maximum cooling efficiency for Intel Core i7 processors and we are going to dwell on it again in the corresponding part of our today’s review.
In conclusion to our today’s discussion of the new Prolimatech Megahalems cooler I would like to add that its recommended price is comparatively high. Besides, the cooler is still quite rare in the market, which should also be considered as a drawback.
In this part of our review we are going to briefly refresh your memory about the today’s testing participants, which all should already know very well by now. We are also going to talk about the peculiarities of these coolers installation onto the LGA1366 platform, since we haven’t talked about it before (because the testbed was not yet available). This is what we will discuss in detail today.
All coolers were tested in random order.
We tested the new model from the Korean Zalman Company in the end of 2008 and our impressions were quite ambiguous. On the one hand, Zalman CNPS9900 LED proved more efficient than its predecessor – Zalman CNPS9700 NT, but on the other, it was still losing to the best super-coolers and was selling at a higher price of $89.90.
Anyway, CNPS9900 LED is currently the best CPU cooler from Zalman, so we couldn’t leave it out of our today’s massive test session.
The LGA1366 retention kit is included into the standard accessories bundle that comes with Zalman CNPS9900 LED. This is what it looks like:
There is no backplate in this kit. Instead the retention frame is held with four screws with plastic washers beneath them that are inserted at the back of the PCB:
Since the default socket retention on the LGA1366 mainboards already has a backplate installed, there is no need for a second one.
The cooler is pressed against the CPU with a retention frame with four screws, just like during installation onto LGA775 platforms:
Zalman CNPS9900 LED looks very beautiful inside the system case:
You may have noticed that the plastic shroud covering the fan has been removed according to the recommendations in the manual (which we didn’t notice last time we tested Zalman CNPS9900 LED). Our tests revealed that removing the shroud has exact same effect as covering the slits in it with some sticky tape, namely, the CPU temperature will drop by 2°C. So, it doesn’t matter if you remove the shroud or tape over the slits – the effect will most likely be the same.
The NH-U12 model from the Austrian Noctua Company is remarkable for longevity in the cooling solutions market, because it first appeared in 2006. Two years later they equipped the same heatsink with a higher-performing fan and new retention mechanism and added “P” to the model name (NH-U12P). Finally, by the end of last year Noctua released the third cooler revision called SE1366, which is designed only for the new LGA1366 platforms.
The package size, shape and design remained exactly the same:
The accessories bundle includes everything necessary to install the new cooler onto the LGA1366 platform. There is also highly efficient Noctua NT-H1 thermal grease, fan resistors, four wire clips for the fans and a Noctua logo sticker:
Noctua NH-U12P is a tower-type heatsink made of 36 aluminum plates, each 0.45mm thick and spaced out at a 2.8mm distance from one another. The plates sit on four copper nickel-plated heatpipes, 6mm in diameter that go through copper nickel-plated base:
As you see, there are two Noctua NF-P12 fans attached to the cooler heatsink and it is not something we have done ourselves. Now the cooler comes bundled with two fans:
Their rotation speed varies between 900-1300RPM and can be adjusted using bundled resistors (the blue one – 900RPM, black one – 1100RPM, none – 1300RPM).
The cooler base is not really crooked, but it cannot be called even, either, because all four corners are dropped a little bit:
Of course, it did affect the quality of the contact between the CPU heat-spreader and the cooler base, as you can see from the thermal compound imprint test:
As you can see, the cooler base is a little protuberant (we evened out the processor heat-spreader using glass surface as a reference, which we are going to talk about later in this article). By the way, since the die size of the Intel Core i7 processors has increased compared with the previous CPU generation, the evenness of the processor cooler base plate will now be of much greater importance for the cooling efficiency than before.
Noctua NH-U12P is fairly easy to install. Its retention is very reliable and provides secure contact between the cooler and the CPU heat-spreader:
The metal backplate should be mounted right against the default backplate on the back of the mainboard:
The cooler is very compact the base and doesn’t interfere with any electronic components around the processor socket:
There is a 55mm gap between the mainboard PCB and the lower heatsink plate. This is what Noctua NH-U12P looks like inside the system case:
This cooler is 4-5°C more effective when it is installed with the heatsink parallel to the PCI-Express slots (the heatpipes will go across the CPU) that is why during our today’s test session e installed it the way shown on the photo on the right.
The recommended price for the Noctua NH-U12P is $64.90, which is exactly the same as before. I think it is good news, because it comes with two fans and pretty expensive thermal grease and the price didn’t get any higher. However, let’s wait until the results of our cooling efficiency tests come in and then draw final conclusions.
Cooler Master V10 is another new super-cooler that we have already tested just a little while back.
It comes bundled with everything necessary to be installed onto LGA1366 platform. Just attach the necessary retention to the base of the cooler…
… and tighten the screw-nuts at the bottom of the PCB over the backplate:
Cooler Master V10 fit into our new system case even without removing the casing. However, once the mainboard with Cooler Master V10 in it was put into our Antec Twelve Hundred system case, it was extremely hard to connect the 24-pin power cable, because this connect is located very close to the horizontal heatsink array of the V10 cooler:
Other than that there were no more problems of any kind during installation. Since we have already discussed Cooler Master V10 in detail in our previous review, there is nothing we could add at this point.
Ultra-120 eXtreme cooler from Thermalright has long been one of the leaders among contemporary cooling solutions and we have tested it quite a few times already. The modification supporting LGA1366 platform comes in a clear plastic package that differs tremendously from the traditional cardboard boxes from this manufacturer:
Now the cooler model name has “1366” added to it. It indicates that the cooler is compatible with the new Intel platform for Core i7 processors. The cooler is, in fact, intended only for the LGA1366 platform, because there are no other retention kits bundled with it:
So, if you wish to us it on any other platform, you will have to purchase the necessary retention kits separately.
Let’s take a look at the cooler:
As you can see, now it comes with a fan in a plastic frame that catches on to the sides of the heatsink with special hooks. Thermalright TR-FDB-1600 fan is built using Fluid Dynamic Bearing technology with promised 60,000 hours MTBF:
The fan rotation speed is constant and equals 1600RPM. The generated airflow is 63.7CFM and the noise shouldn’t exceed 28dBA. The cooler with a fan attached to it weighs 946g.
Our cooler sample was not free from the issue typical of almost all CPU coolers from Thermalright, namely, prominent base:
Note that one of the corners of the base plate is dropped more than the other three. The thermal compound imprint on the CPU heat-spreader speaks for itself:
The cooler is installed onto LGA1366 mainboards with the help of an X-shaped plate that should be put over the base and tightened with included screws to the backplate:
Besides Thermalright Ultra-120 eXtreme-1366 RT, we also have an all-copper version of it in our lab called Thermalright TRUE Copper Ultra-120 eXtreme:
The only differences between these two models are the heatsink material and a significantly larger weight of the copper cooler (it increased from 790g to 1900g). The base is even, which we have already pointed out in the dedicated cooler review.
Both modifications of Thermalright Ultra-120 eXtreme cooler look the same when they are installed into the system case, so I will only provide the photo of the copper version here:
I would like to add that during our test session we used two plastic ties attached to the top case panel to hold the copper Thermalright Ultra-120 eXtreme. Just like Prolimatech Megahalems, both Ultra-120 eXtreme coolers proved more efficient when they were installed with the heatpipes going across the processor heat-spreader.
This is the newest and so far the best cooler with the most attractive set of features for its price. We have reviewed the new Scythe Mugen 2 cooler just recently:
In fact the cooler retention for the LGA1366 platforms doesn’t actually differ from the one for LGA775 mainboards. It consists of two plates screw on to the cooler base:
The cooler is fastened to the board with a universal backplate replacing the default one. There is a special wrench included with the cooler accessories that you should use to remove the default backplate. By the way, those who are going to purchase this cooler should first consult the Scythe Mugen 2 compatibility list to make sure that it will be compatible with the mainboard.
The cooler base is ideally even that is why we are going to talk about the thermal compound imprints left by the base of this cooler on the processor heat-spreader in a separate part of our today’s article.
Take a look at Scythe Mugen 2 inside the system case:
Unfortunately, if the cooler is installed in a different position (turned by 90°C), its fan blocks the first and second memory DIMM slots completely. Therefore, during our test session the cooler was installed the way shown on the photo above (with the heatpipes directed along the processor heat-spreader and the airflow sent towards the top 200-mm exhaust fan).
Another highly efficient cooler that will participate in our today’s test session is ThermoLab BARAM:
The sample we got didn’t have the retention kit for LGA1366 platform included with the bundled accessories that is why they sent it to us separately:
A small plastic bag contained a universal backplate, installation instructions, washers, two sticky pads of different thickness, retention panels and screws:
The entire installation procedure for the LGA1366 mainboards requires attaching the retention panels to the base of the cooler and fastening them on the board with screws tightened to the backplate through the PCB:
However, there is one very important peculiarity. The thing is that the metal backplate is not flat – its corners are a little curved to one side. It means that if you install the backplate not the way the manual advises, but with the backplate corners bent away from the PCB, the cooler will be pressed harder against the processor heat-spreader, which should ensure much better heat transfer. Moreover, do not forget to stick a thin pad to that part of the backplate that will be pressed against the original mainboard backplate. This will add a few extra millimeters to the distance between the backplate and the board and thus increase the pressure even more. Air-cooling maniacs may use a thick pad instead; however, you will need to push hard against the cooler during installation in this case.
This is what ThermoLab BARAM looks like installed into a system case:
Overall, there is nothing else I could add here, so let’s move on to the next candidate.
The most effective cooler of so-called top-type (when the airflow is directed towards the mainboard) – Thermalright AXP-140 – has also been tested in our lab before.
Since there is no LGA1366 retention kit among the bundled accessories, but the cooler’s base size and shape are identical to those of the base on Thermalright Ultra-120 eXtreme, we decided to us the retention kit from the latter:
Of course, it was not too convenient to tighten the screws, but the pressure was high enough, which produced a very good thermal compound imprint:
During our tests the cooler was equipped with a Scythe Kaze Maru fan, which is the most suitable fan for AXP-140 providing the highest cooling efficiency:
I would like to add that the AXP-140 cooler installed onto the board blocked the first memory DIMM slot and touched the tall heat-spreader on the memory module installed into the second slot.
Thermalright IFX-14 cooler has been considered the ultimate best CPU cooler. Its LGA1366 compatible model comes in a completely new packaging:
The plastic casing with flames all over it is extremely appealing and beautiful compared with the old boring cardboard boxes. The cooler is accompanied with the corresponding accessories for proper installation onto LGA775 and LGA1366 platform. There are also thermal grease and installation instructions in several languages:
Thermalright IFX-14 is based on four copper nickel-plated heatpipes, each 8mm in diameter. They hold two independent arrays of aluminum heatsink plates:
There are a total of 108 plates. The space between them measures 1.5mm, and each is 0.25mm thick. The calculated effective heatsink surface measures 10,323sq.mm and so far it is one of the best in our today’s test session.
Both IFX-14 models that we have in our lab have a prominent base, just like almost all Thermalright coolers:
At first we had one of the earliest IFX-14 samples that didn’t have retention for LGA1366, that is why we used the retention kit from Prolimatech Megahalems cooler. To ensure secure contact with the processor heat-spreader and prevent the cooler from shifting sideways, we glued a 5-mm thick piece of rubber to the retention plate:
Moreover, the plate itself was taped over to prevent it from sliding during installation and the springs on the screws were replaced with metal and rubber washers:
As a result, we ended up with a very reliable retention with strong hold. If you try to use the default Prolimatech Megahalems retention kit without any modifications of your own, IFX-14 cooler will not be pressed securely against the processor heat-spreader and will shift under its own weight.
The new Thermalright IFX-14 cooler that we received right before completing the tests already has LGA1366 retention of its own. It turned out very similar to what we put together ourselves:
Moreover, the comparative testing of the Thermalright IFX-14 cooling efficiency with two different types of retention (hand-made and original) didn’t reveal any difference in the CPU thermal readings. Nevertheless, we decided to use the original retention kit for our primary test session.
Thermalright IFX-14 cools the CPU 3°C better if its heatpipes are positioned horizontally (we are talking about tower cases here). This is what it looks like:
You can top the cooler with one or two 120x25(38)mm or 140x25(38)mm fans. So, today we are going to test it in all these configurations.
Thermaltake BigTyp 14Pro is currently the most high-performance cooling solution from Thermaltake. It has recently replaced Big Typhoon 120 and is now among respected super coolers for LGA 775 platforms.
Unfortunately, BigTyp 14Pro isn’t bundled with the LGA1366 retention kit that is why we had to use one from Thermalright Ultra-120 eXtreme-1366 RT, just like we did for Thermalright AXP-140. However, since the base of Thermaltake BigTyp 14Pro is thinner than that of Thermalright coolers, we put a 5-mm thick piece of rubber underneath the crosspiece of the retention:
As a result, the cooler was pressed just perfectly against the processor heat-spreader, although it was really hard to achieve this kind of secure contact (the bent retention plates show how hard we had to press it).
Since we had to use the retention that had been originally designed for a different cooler, the base wasn’t strictly in the center of the CPU heat-spreader, because of the angled heatpipes coming out of the cooler base. They hit against the X-shaped retention plate. Therefore, the thermal compound imprint on the cooler base looked as follows:
As you see, one heatpipe is barely taking part in the heat transfer, which makes proper testing of this cooler impossible. Nevertheless, we decided not to give up tests and would like to wish Thermaltake to release their own LGA1366 retention kit for the BigTyp 14Pro cooler. The interesting thing is that they already have the new retention kits for V1/V1 AX coolers, but not for the flagship Thermaltake BigTyp 14Pro solution.
Let’s take a look at the cooler installed into the system case:
The retention limitation causes the cooler to slightly turn along the horizontal axis.
The second cooler from Cooler Master has also recently joined the best air coolers. It is called Cooler Master V8:
It is made of four independent heatsink arrays, 8 copper nickel-plated heatpipes and a 120-mm fan with PWM rotation speed control. We used the LGA1366 retention kit from Cooler Master V10 that was a perfect fit. As a result, it produced a very good thermal compound imprint on the cooler base:
Cooler Master V8 is comparatively compact when installed into a system case.
Nothing can prevent you from installing it any way you want with the airflow sent in any direction. As it turned out during our test session, when the cooler is installed with the heatpipes going across the processor heat-spreader and the airflow directed towards the back of the case (and not towards the top, as you see on the photos above), the cooling efficiency increases by 3-4°C. Therefore, we decided to run all the tests with the cooler installed in this particular way.
The technical specifications of all cooling solutions participating in our today’s test session are summed up in the table below:
The next diagram shows the calculated effective surface of the coolers heatsinks…
… and the next one – their price:
Our testbed was identical for all coolers throughout the test session and featured the following configuration:
All tests were performed under Windows Vista Ultimate Edition x86 SP1. We used the following software during our test session:
So, the complete screenshot during the test session looks as follows:
The stabilization period for the CPU temperature between the two test cycles was about 10 minutes. We took the maximum temperature of the hottest processor core of the four for the results charts. The ambient temperature was checked next to the system case with an electronic thermometer with 0.1°C precision that allows monitoring the temperature changes over the past 6 hours. During our test session room temperature stayed at 23.5-24°C.
The noise level of each cooler was measured after 1:00AM in a closed room about 20sq.m big using CENTER-321 electronic noise meter. The measurements were taken at 3cm, 1m and 3m distance from the closed system case. During the acoustics tests all five 120-mm case fans were slowed down to ~520 RPM. In this mode the background noise from the system case measured at 1m distance didn’t exceed ~33.3 dBA. When the system was completely powered off, our noise meter detected 30.8 dBA (the lowest on the charts is 30 dBA). The subjectively comfortable noise level is around 34.5~35 dBA.
Before we moved on to the actual testing of the cooling solutions performance we decided to study the technical specifications and physical parameters of a very important component of a CPU that may affect the cooling efficiency quite significantly. I am talking about the processor heat-spreader. First let’s talk about its technical specs. Our Intel Core i7-920 processor was made in Costa-Rica and marked as “SLBCH”:
Its nominal frequency is 2667MHz and core voltage – 1.25V. Other specs are as follows:
We tested the evenness of the heat-spreader using a metal ruler and it seemed OK:
You can notice a slight bump in the CPU cross-section, which showed in the thermal compound imprint on the ideally even polished vase of the Scythe Mugen 2 cooler:
Of course, an imperfect imprint like that is totally unacceptable for our today’s massive test session that will determine the fate of quite a few cooling solutions. That is why we decided to void the warranty and even out the heat-spreader.
After a few minutes of work on a glass with fine sanding paper we could clearly see the unevenness:
After another 15 minutes and one more even finer sanding surface, we managed to eliminate all the major bumps:
Fixing the CPU heat-spreader surface as described above immediately lowered the temperature of the hottest processor core by 4°C! Impressive, isn’t it? However, when we removed the CPU cooler, the thermal paste imprint didn’t impress us at all:
Another 15 minutes of work on the glass surface with the finest sanding paper we could find improved the CPU temperature by another 1-2°C and generated a more or less acceptable imprint:
We decided to skip the polishing of the heat-spreader to mirror-shine. And the reason for that is explained very well on the scheme from the official Intel web-site:
Look at the actual die size beneath the heat-spreader: it is only 14.3 x 18.8 mm, which makes 268.84sq.mm (256sq.mm according to Intel’s official specs). So, the die size makes only a little over 20% of the heat-spreader. Therefore I believe that the main heat flow will be going from the CPU to the cooler in the spot right above the actual die. That is why the main goal is goal is to ensure a perfect contact in the center of the heat-spreader. That is exactly why Thermalright coolers are so efficient even despite their prominent base plates. Of course, if you even out the base on these coolers you will improve the CPU temperature a little, because some heat is still distributing over the CPU heat-spreader.
And one more thing that we should dwell on in this part of our today’s article. Look at the CPU core under the heat-spreader:
It is evident that the die is of rectangular shape and its longer side goes along the processor heat-spreader. In other words, when you install the CPU into the board, the die will be positioned along the longer side of the board. It is logical to assume that the more heatpipes will fall into the die zone, the more efficiently the cooling will be. So, the coolers should be installed onto Core i7 processors with their heatpipes going across the CPU (parallel to PCI-E slots on the board). We paid special attention to this particular matter in our today’s review on purpose. So, if we analyze everything we know, we will see that the majority of coolers that could be installed with their heatpipes across the processor did, in fact, prove more efficient than with the heatpipes along the CPU. The only exception was Thermalright IFX-14, however, it could be its uneven base that contributed to the final result. Until it is evened out, we can’t really make any conclusions about the preferable cooler orientation on the CPU.
In conclusion to this part of our article I would like to offer a small diagram with the Intel Core i7 thermal readings under workload created by different applications:
Just like with Intel Core 2 Quad/Duo, the highest processor load is created by Linpack algorithm. OCCT program may have an error (or may be using a different Linpack version), because it heats the CPU less than LinX.
First of all, let’s check out the cooling efficiency of the new Prolimatech Megahalems solution depending on the rotation speed and number of fans. We used Scythe Slip Stream 120 fans at 500-1920RPM rotation speeds. They differ from other 120-mm fans by the rotor of smaller diameter and a strong airflow in the entire speed range. The fans were installed onto Prolimatech Megahalems cooler for intake and exhaust with the airflow directed towards the top of the system case:
We tested the efficiency of the new cooler in the following manner. First we determined maximum CPU overclocking with only one fan installed and working at 500RPM. Then we added a second fan with the same rotation speed, tested the CPU at the same frequency and then again determined the maximum overclocking but this time with two fans. The next step was to run the exact same tests with one fan rotating at 350RPM higher speed, and so on and so forth. The results are summed up on the diagram below:
The results show that the cooling efficiency of Prolimatech Megahalems cooler depends a lot on the fans rotation speed, up until the maximum speed. In other words, the claims of Prolimatech engineers about the Megahalems heatsink being optimized for fans with 800-1200RPM rotation speeds are not quite correct. With an additional fan installed for air exhaust or with the rotation speed of a single fan increased, we can see the CPU temperature go down by 4-7°C.
The overall CPU frequency gain that you get by switching from a single 500RPM fan to two fans at 1920RPM made +218MHz accompanied by the 7°C temperature drop on the hottest CPU core. However, if you take a closer look at the diagram, you will see that the most frequency increase occurs when we install two fans and raise their rotation speed to 860RPM. Further rotation speed increase doesn’t justify the significantly growing level of noise.
Now comes the most interesting part of our test session, that part that took us over two weeks of tests and over 4 days of editorial work. When we ran Linpack tests our quad-core 45nm CPU overclocked to 3.76GHz (+41%) on the weakest cooling system of all participating. The processor core voltage was increase only by 0.075V to 1.275V (+6.3%) in the mainboard BIOS.
We activated “Load-Line Calibration” that lowers the voltage drop on the segment between the voltage regulator and the CPU itself. The memory voltage was set at a fixed value of 1.525V, its frequency equaled 1500MHz with 8-8-8-18 timings:
All other parameters in the mainboard BIOS connected with the CPU and memory overclocking remained unchanged (left at “Auto”).
In this part of our session all coolers were tested with their default fans. If the cooler was equipped with a rotation speed controller, it was tested in two modes: at min and max fan rotation speed. Cooler equipped with PWM-controlled fans were tested in Auto mode (“Turbo” profile in the mainboard BIOS). Since Thermalright TRUE Copper ships without a fan, we tested it with a fan from its fellow cooler – Thermalright Ultra-120 eXtreme-1366 RT – fan model TR-FDB-1600. The same is true for the Thermalright IFX-14 cooler, which also ships without a fan. The tower-heatsinks – Prolimatech Megahalems and ThermoLab BARAM – were topped with a fan from Scythe Mugen 2 with the rotation speed varying between 250 and 1300RPM. The last cooler that has no fan with it, Thermalright AXP-140, was tested with the most suitable fan – Scythe Kaze Maru at 1150 and 1920RPM rotation speeds. I would like to stress once again that we used the same thermal grease for all coolers: it was Arctic Silver 5. The case was also the same: Antec Twelve Hundred.
The red carpet and drum roll are on. Let’s check out the results now:
The coolers on the diagram are lined up in order of efficiency starting with the most efficient one at the very top. Let’s try analyzing the obtained results. The leader is Thermalright IFX-14 with a 4°C advantage over the competitors, which is quite a lot. Overall, when we tested Thermalright IFX-14 we got the impression that this cooler was preeminent among the others. Further tests will confirm this observation.
The leader is followed by a dense group of 11 coolers from Scythe Mugen 2 to Thermalright AXP-140 (at 1920RPM). The difference between the best and the worst in this group is 4°C. The best ones here are Scythe Mugen 2 and Prolimatech Megahalems; only one degree behind them are ThemoLab BARAM and Thermalright Ultra-120 eXtreme that cool the CPU equally well (in this test mode). Noctua NH-U12P SE1366 with two fans at maximum and medium rotation speeds loses another degree. The next one is again Noctua NH-U12P SE1366 at the slowest fan rotation speed, Thermaltake BigTyp 14Pro at 1050RPM and Cooler Master V8. The 11-cooler group is closed by Thermalright AXP-140 with a fan at 1920RPM that yielded to the cooler of the same type – Thermaltake BigTyp 14Pro.
The remaining coolers fall into the outsider group, however, it is important to remember that they are the least efficient coolers of the absolute best out there. Here we see Thermaltake BigTyp 14Pro and Thermalright AXP-140 at the minimal rotation speed of their fans, Cooler Master V10 that has suddenly yielded to the V8 model on Intel Core i7 platform. The very last one in this race is Zalman CNPS9900 LED, which has limited the maximum CPU overclocking for this test. And since we know the maximum CPU overclocking result achieved on the weakest cooler of the today’s testing participants, it is time to check out what the overclocking maximums will be for the rest of the pack:
The difference between maximum overclocking results for each of the cooling solutions is not dramatic, and there is simply no difference between the coolers within their efficiency group. Nevertheless, we can notice a difference in the CPU temperatures at the same frequency and voltage settings. The indisputable leader here is again Thermalright IFX-14 winning 5°C from its closest rival, Scythe Mugen 2. The latter is closely followed by Noctua NH-U12P SE1366 with two fans and two Thermalright Ultra-120 eXtreme coolers. The all-copper TRUE model has finally proved worthy having outperformed its aluminum brother by 2°C. By the way, to ensure CPU stability when tested with Ultra-120 eXtreme with aluminum heatsink, we had to increase the processor core voltage to the next level. At the same time voltage increase didn’t help improve the Core i7 stability at 3958MHz with ThermoLab BARAM cooler, so we had to set the base frequency 1 step lower. Please, do not forget that we are testing these coolers with their default fans, so Thermalright coolers have a slight advantage thanks to a slightly faster fan (although we could argue about it, because the 9-blade Scythe Slip Stream at 1300RPM will hardly yield to the 7-blade Thermalright fan at 1560RPM in terms of the created airflow strength).
After that we see gradual lowering of the stable CPU frequency and increase in its core temperature. The least successful ones here are Cooler Master V10 and Zalman CNPS9900 LED. The new Prolimatech Megahalems rests modestly somewhere in the middle losing in efficiency to the ultimate cooling masters.
Well, now that we have arranged the coolers in order of efficiency when tested in their default configurations (or equipped with related fans), it is time to run the tests with identical fans and find the most efficient heatsink among our today’s testing participants. However, since there are way too many coolers in our race, we decided to test only six best of them here.
We chose Silverstone FM123 fans for our next test:
These fans have a rotation speed controller: their rotation speed may be adjusted from 900 (±90) to 2600 (±260) RPM creating 40.7-106.3CFM airflow and 0.66-4.55mmH2O static pressure. The fans were installed for intake-exhaust with the airflow directed towards the top of the case:
Since it is not very smart to run the tests at 2600RPM fan rotation speed with the closed system case, we removed the side panel for the entire test session:
Unlike all other coolers, Thermalright IFX-14 can accommodate not just one or two but three fans at the same time, and these can be not just 120x25(38)mm fans, but also 140x25mm fans. Since we didn’t have three identical Silverstone FM123 fans at our disposal, we installed three 140-mm Scythe Kaze Maru fans on it.
Quite a monster, isn’t it? All fans worked at 1260RPM rotation speed. Now let’s check out the obtained results:
Here is the screenshot of the best test result:
Once you check out the results, you ask yourself: if we call all 13 participating cooling solutions “super coolers”, then what shall we call Thermalright IFX-14? An “ultra cooler”? The CPU core voltage and frequency are higher, while the temperature is still lower than by other five contestants. Among them we would like to point out Scythe Mugen 2 and ThermoLab BARAM as offering the best feature and efficiency combination. Other coolers, however, also didn’t fall too far behind.
The results of our acoustic measurements are given on the diagram below:
The subjective comfort zone embraces five coolers including Cooler Master V8 at its minimal fan rotation speed. Then come coolers and fans with medium noise: this group starts with Thermalright AXP-140 at 1150RPM and finishes with Noctua NH-U12P SE1366 with two fans at 1380RPM. All other testing participants turned out pretty noisy, so you will hardly be able to stay around them for a long time without feeling any discomfort.
Well, it’s time to sum thing up.
Today’s test session didn’t reveal anything sensational: the most efficient cooling system is again Thermalright IFX-14. Its advantage over the rivals on the Intel Core i7 platform is indisputable. This cooler left everyone behind in both configurations – with one and two fans. As for the configuration when IFX-14 comes with three 140-mm fans, only high-quality liquid-cooling systems could probably try to beat that. The drawbacks of this remarkable solution are high price, no fans among the bundled accessories and traditionally uneven base. If we disregard all that, IFX-14 will be an absolute winner. Therefore, we are proud to crown Thermalright IFX-14 with our prestigious Editor’s Choice title for being the ultimate best cooling solution:
Among the highly efficient coolers from a dense group following the leader we should single out Scythe Mugen 2. This cooler not just outperforms its direct rivals if we consider the total score from all the tests, but it is also the cheapest of all! Mugen 2 comes bundled with a high-performance fan that is inaudible in quiet mode and produces moderate noise under high CPU workload. Moreover, it is a universal cooler, unlike Thermalright Ultra-120 eXtreme-1366 RT, Noctua NH-U12P SE1366 and Prolimatech Megahalems. Taking into account that Scythe has very well-established dealer network, things will become very tough for their competitors.
Among the cooling solutions that disappointed us with their performance on Core i7 platform we should mention Cooler Master V10 and Zalman CNPS9900 LED. The former is not only expensive, but is also extremely large in size that makes it difficult to install and remove it. Moreover, it may not even fit in some cases at all. The solution from the Korean manufacturer that performed very well on the 45nm Core 2 Quad processor, seems to be lacking heatpipes (there are only three of them) and effective heatsink surface. I really think it is time for Zalman to finally launch something impressive not only in terms of pricing, but also efficiency.
Well, that is about all for today. We hope that this article will help you decide on a cooler for your CPU. And as for us, we are going to continue our study of super-coolers performance, because by the time the article was completed, we got our hands on three more ambitious candidates, which we are going to introduce to you shortly. So, stay tuned for more hot cool stuff!