by Sergey Lepilov
07/07/2010 | 11:59 AM
When you are buying a new more powerful processor to replace your old one, you often ask yourself if a cooling system upgrade is also necessary. And while this question is not very difficult to answer for the mainstream user, because all contemporary cooling systems worth $25 or higher can in fact cope just fine with any CPU out there in its nominal operation mode, then overclocking fans should really think this matter through thoroughly. Very often it is the cooler that limits overclocking of processor from the upper price segment, which have an unlocked clock frequency multiplier. So, today we are going to check out six processor coolers using our new six-core Intel Core i7 extreme Edition i7-980X processor and will see how its overclocking success depends on the cooler choice. And since four coolers out of these six are new to us, let’s talk about them in a bit more detail first.
The new creation of the well-known Japanese Scythe Co., Ltd. is called Yasya and is shipped in a medium sized box with the cooler photo on the front:
As usual, Scythe’s packaging is completely covered in all sorts of information starting with the list of supported platforms and finishing with warranty details:
There is a smaller box at the bottom of the main package, which contains two pairs of retentions, a manual, a pack of SilMORE thermal paste and two wire clips for the fan:
Scythe Yasya tower cooler is made in Taiwan. It consists of a copper nickel-plated base with six copper heatpipes 6 mm in diameter soldered to it. They hold aluminum heatsink plates:
The cooler with a fan attached to it weighs no more than 848 g. it measures 130x108.5x159 mm. the two schematic images below show more details pertaining to the dimensions of this cooler:
The main peculiarity of Scythe Yasya cooler is wedge-shaped plates that are shifted away from one another:
The Japanese engineers called this technology “Trident Multi Layer Fin Structure”. It is supposed to reduce the fan airflow resistance and therefore increase the heatsink cooling efficiency. We have already seen many solutions like that, but Scythe Yasya seems to have it implemented in the most radical way. I also have to add that all plates are polished to mirror-shine. Their impeccably smooth surface should also improve the airflow, which means that Yasya should be pretty effective even with low-speed fans, which should undoubtedly be great news for the owners of quiet systems.
Overall, the heatsink includes 54 plates, each 0.35-0.4 mm thick that alternate with 1.8 mm gaps:
The sides of the heatsink aren’t covered. There I a small aluminum heatsink at the bottom of the cooler, right above the heatpipes.
Another technology utilized in the new Scythe Yasya is called “Unparallel Six Heatpipe Construction”. It implies that the heatpipes are laid out non-linearly inside the heatsink body:
We are also already familiar with the staggered-order heatpipe layout, as other cooler manufacturers use it in their products, too. It allows to distribute the heat more evenly over the heatsink body, speed up the heat transfer and increase the cooler efficiency.
The base of Scythe Yasya cooler is covered with a plastic film that needs to be removed before installation:
Once the film has been removed, you see a very even base with ideally polished surface:
Even the thermal compound imprint left by our processor with a protuberant heat-spreader (we are going to talk more about it a little later in our review) turned out quite OK:
The only remark we have about the base of Scythe Yasya cooler, is that the heatpipes are slightly flattened and soldered to the 2 mm base plate instead of being placed into the grooves.
Scythe Yasya is equipped with one Slip Stream 120 Adjustable PWM (SY1225SL12HPVC) fan:
We are already very well familiar with this model, but now it supports PWM control and comes with a regulator on a bracket for the case rear panel, which can set the maximum fan rotation speed:
According to the specifications, the fan supports the following operational intervals: 470(±30%)–1.340(±10%) and 740(±25%)–1.900(±10%). The claimed airflow is 23.0-76.53 and 37.15-110.31 CFM, and the noise is at 7.05-27.3 and 9.8-37.0 dBA respectively. The actual fan impeller is 112 mm in diameter, and its rotor is 35 mm in diameter. It comes with a 300 mm cable and a 300 mm cable leading to the rotation speed regulator.
The fan is attached to the heatsink using two enclosed wire clips and contacts directly with the plates without any shock-absorbing strips or the like:
It is strange to find only two clips among the bundled accessories, because the symmetrical Yasya cooler can easily accommodate two fans. So why did Scythe decide to save on the wire clips?
The new Scythe Yasya cooler is compatible with all contemporary home PC platforms without exceptions. The cooler is installed onto the CPU and the mainboard without any additional tools and without removing the mainboard from the system case. You need to insert a pair of retention brackets into the sides of the additional heatsink above the base. After that these retention brackets will simply catch on to the plastic frame around the CPU socket on AMD platforms, or will be locked by applying serious pressure from the top on Intel platforms:
The distance from the lower heatsink plate to the cooler base is 35.5 mm, which – with the height of the CPU socket – is quite enough to ensure Yasya’s compatibility with the heatsinks over the mainboard voltage regulator components.
There is no mention of the preferable cooler positioning anywhere in the manual or on the package that is why we decided to check out its efficiency in two positions:
After a few test cycles it turned out that when Scythe Yasya is installed with its heatpipes going across the processor heat-spreader (the photo on the right), the peak core temperature is stably 3-4°C lower than in case the cooler is installed with the heatpipes along the heat-spreader. We believe that the cooling efficiency of six-core processors depends on the way the heatpipes are positioned in respect to their heat-spreader. Especially, since this statement didn’t get confirmed only on one cooler out of six products tested within this session. But, let’s not get ahead of ourselves here.
In conclusion to our Scythe Yasya coverage we have to add that its recommended retail price is $57 and it comes with a one-year warranty.
In the end of April ThermoLab Co., Ltd that launched a very successful Baram cooler about a year and a half ago, decided to refresh this product by introducing a few constructive modifications. Baram-2010 is the new name of the flagship air-cooler by ThermoLab.
The new cooler comes in a not very large box with an image of the cooler on the front and a specification on one of the sides:
I have to remind you that the previous box looked less attractive. The cooler is bundled with a complete set of accessories for Baram-2010 installation on all contemporary platforms:
The new accessories bundle is also a little different from what the first Baram came with: now it includes a universal backplate, retention mounts and rubber rings for them, screw-nuts and a step-by-step installation guide in several languages. Although there were no principal changes in the installation procedure for the new Baram-2010. Thermal compound and four wire clips for the fans didn’t go anywhere. Just like its predecessor, Baram-2010 comes without a fan, so the users have to make their own fan choice. The cooler is made in South Korea and is priced at $50 MSRP.
At first glance it seems like nothing has changed: the same five copper heatpipes 6 mm in diameter spread out widely, the same aluminum plates with wave-shaped edges that are shifted away from one another:
Only the embossed “ThermoLab Baram” words on the heatsink plates make it visually different from the predecessor. Nevertheless, there are a few new things about this cooler. The most important change is the increased number of heatsink plates that is now 64 instead of 54. As a result, the effective heatsink surface size has increased from 7580 to 8900 cm2! And since the heatsink dimensions (132x67x160 mm) and plate thickness (0.45 mm) remained the same, you can easily guess that now the plates density has increased. And this is absolutely correct: the gaps between the heatsink plates have been made smaller: 1.6 mm instead of 2 mm. These heatsink modifications are obviously intended to ensure that medium- and high-speed fans will be used effectively with this heatsink. However, since Baram’s heatsink has been seriously optimized we may encounter some surprises at low fan speed as well. I would also like to add that the heatsink has become heavier and weighs 710 g instead of 625 g, which is still less than the weight of other manufacturers’ flagship solutions.
Besides, now the perfectly finished cooler base is not nickel-plated anymore:
Theoretically, the absence of nickel-plating can also increase the cooling efficiency a little bit more. Moreover, ThermoLab engineers mention higher precision contact between the base, the heatpipes going through it and the plate covering them at the top: they are all soldered together. As you understand, there is no way for us to check whether it is true, but it is extremely important for ensuring fast and even heat transfer from the processor heat-spreader to the cooler base.
The cooler base is impeccably flat, and the uneven imprint left by the thermal compound can be explained by the protuberant heat-spreader of our test processor, just like with Scythe Yasya:
ThermoLab Baram-2010 is compatible with all contemporary platforms. The cooler is fastened through the mainboard PCB using a universal retention plate. The retention of the new cooler has been modified to allow tool-free installation. Baram-2010 is tightened up using large screw-nuts:
Here ThermoLab specifically points out that the cooler is pressed against the processor with 20 kg of force, which is less than by other manufacturers’ solutions (45-62 kg), but sufficient for effective heat transfer. When you tighten the screw-nuts, you feel like they could be tightened a little more. The distance between the lowest heatsink plate and the mainboard PCB is 45 mm.
As we have already said, Baram-2010 comes with no fans, but you can attach the fans to the heatsink using two wire clips. There are no shock-absorbing strips included with the cooler.
This is all I could tell you about the new Baram-2010 before we get to test its cooling efficiency. Let’s move on to the next cooler.
The next participants of our today’s test session is made by a well-known company called Spire. The cooled we are going to talk about is TherMax Eclipse II (SP984B1-V2). It ships in a large cardboard box with a cut out window in the front:
However, this window covered with clear plastic is too small to let you see the new cooler, so you will need to take it out of the box in order to check it out better. The box bears detailed information about the cooler and the platforms it supports. TherMax Eclipse II is bundled with the following accessories:
When we looked at the retention kits included with the cooler, we got the familiar feeling of déjà vu, which actually proved perfectly normal under the circumstances:
I am sure that the heatsink of the new TherMax Eclipse II reminds you a lot of the recently tested Alpenföhn Nordwand. However, TherMax Eclipse II stands out due to two fans which are attached using eight silicone mounts:
The base with heat-pipe direct touch technology is identical to that of the above mentioned Alpenföhn cooler:
It is finished off quite well and is pretty flat that is why the responsibility for uneven thermal compound imprint lies solely on the protuberant CPU heat-spreader:
The rotors of two 120 mm fans have a Spire sticker, although they look very similar to Cooler Master fans:
Surprisingly, Spire engineers decided not to bother with implementing fan rotation speed adjustment option: the fan rotates with constant speed of 2200 RPM. There is no PWM control of any kind, so the users will have to connect the fans to the fan rotation speed controller or put up with the noise from a pair of 120 mm fans working at such high speeds. However, some mainboards can adjust the rotation speed of the fans connected to them, so if you own one of those boards, you will be able to set TherMax Eclipse II fans at a more comfortable acoustic level.
The cooler is compatible with all contemporary platforms and we have already described the installation procedure in detail many times.
Spire TherMax Eclipse II is priced at $52.99, which is quite acceptable for a cooler with two good fans.
He fourth participant of our today’s test session will be the so-called “boxed” cooler for the Intel Core i7-980X Extreme Edition processor. It is also accompanied with a small instructions sheet and a syringe with thick gray thermal compound.
Although we see all boxed coolers as a chunk of aluminum with a copper center and a buzzing fan, the new cooler looks very decent. It is a tower cooler with four copper heatpipes 6 mm in diameter:
The cooler measures 96 x 100 x 125 mm and weighs about 870 g. The design of the cooler is very illustratively depicted on the following schematic chart:
The heatsink array consists of 57 aluminum fins measuring 100x70 mm that are 0.35 mm thick:
Small inter-plate gaps of only 1.1~1.2 mm and no optimizations of any kind aimed at lowering the airflow resistance indicate potentially high dependence of the cooler efficiency on the fan rotation speed. At the same time, the heatsink is topped with a frameless fan that is way worse-suited for a heatsink of this density that fans that have a frame to them.
Heatpipes are soldered to the plates and the copper base that is perfectly finished:
The base surface is extremely flat, no complaints here, unlike the processor heat-spreader:
The rotation speed of the 100 mm fan is PWM controlled in two ranges, set using a small switch on top of the cooler:
In Quiet mode (“Q”) the speed is adjusted in the interval between 800 and 1800 W, and in maximum speed or performance mode (“P”) the supported range extends from 800 up to 2600 RPM. We don’t know what the declared airflow is, and the noise level should vary between 17 and 35 dBA for the corresponding speed ranges. The fan bearing should last 57,000 hours.
The cooler is only fit for the LGA1366 platform. It is installed using a backplate through the mainboard PCB and provides very high-pressure hold.
There are only 33 mm between the lowest heatsink plate and the cooler base, but the cooler is extremely compact, so it won’t interfere with any of the heatsinks on the mainboard voltage regulator components as well as with the tall memory heat-spreaders.
The fan features blue LEDs, which brightness depends on the fan rotation speed:
The cooler is priced at about $50, but we do not know whether they will be selling it separately, like the boxed cooler for LGA1156 platform.
Our new test CPU is from the top price segment for home system CPUs. Its marking reads “SLBUZ”. We have already talked in detail about this processor in our article called “Six Cores for a Desktop: Intel Core i7-980X Extreme Edition CPU Review”.
This processor sample features a single copper nickel-plated heat-spreader of protuberant shape:
Note that the bump is not really in the center of the heat-spreader, but has been shifted closer to one of the sides. About two thirds of the heat-spreader along the shorter side of it is quite flat, while one third is obviously going south. As for the longer side of the heat-spreader, there are no evenness issues here, as we can tell from applying our available tools. You have already seen what imprints heat-spreaders like that leave on the cooler bases, and you are going to see two more a little later in this review. Despite everything we have just said, we are not going to lap the heat-spreader this time: we will do it in one of our upcoming reviews.
The die of the Extreme Edition processor made with 32 nm process measures 14.3x18.8 mm and is positioned horizontally to the heat-spreader, just like the die of regular Intel Core i7 CPUs:
Therefore, we can assume that the cooling efficiency will depend on the way the cooler heatpipes are positioned in reference to the processor heat-spreader. However, we will need to lap the CPU heat-spreader flat in order to confirm or deny this assumption.
These are the specifications of our Intel Core i7-980X Extreme Edition processor:
CPU-Z caught Turbo Boost technology in action that is why you can see a slightly different frequency than the nominal one:
The nominal processor Vcore for our unit is 1.225 V.
We are going to test the cooling efficiency of our today’s testing participants in the following closed testbed:
Processor overclocking was limited by the least efficient cooler of our today’s testing participants in its quiet mode. As a result, we managed to overclock our six-core processor (with protuberant heat-spreader) with the multiplier set at 23x and “Load-Line Calibration” enabled to 4.3 GHz. The nominal processor Vcore was increased to 1.3975 V in the mainboard BIOS:
Besides, we manually set the following voltages in the mainboard BIOS:
Turbo Boost and Hyper-Threading technologies were disabled during our test session. The memory voltage was at 1.64 V and its frequency was around 1.53 GHz (7-7-7-14_1T timings). All other parameters available in the mainboard BIOS and connected with CPU or memory overclocking remained unchanged.
All tests were performed under Windows 7 Ultimate x64 operating system. We used the following software during our test session:
So, the complete screenshot during the test session looks as follows:
The CPU was loaded with two consecutive Linpack test runs with the settings as indicated above. The stabilization period for the CPU temperature between the two test cycles was about 8-10 minutes. We took the maximum temperature of the hottest CPU core for the results charts. Moreover, we will also provide a table with the temperature readings for all cores including their average values. 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. The room temperature during our test session varied between 22.9-23.2 °C.
The noise level of each cooler was measured between 1:00 and 3:00 AM in a closed room about 20 m2 big using CENTER-321 electronic noise meter. The noise level for each cooler was tested outside the system case when the only noise sources in the lab were the cooler and its fan. The noise meter was installed on a tripod and was always at a 150 mm distance from the cooler fan rotor. The tested cooling systems were placed at the edge of the desk on a sheet of polyurethane foam. The lowest noise reading our noise meter device can register is 29.8 dBA and the subjectively comfortable noise level in these testing conditions was around 34 dBA (do not mix it up with low noise level). The fan(s) rotation speed was adjusted in the entire supported range using the new controller revision by changing the voltage with 0.5 V increment.
We are going to compare our newcomer against the performance reference – Noctua NH-D14 air-cooler equipped with two 140 mm Noctua NF-P14 fans:
We tested the super-cooler in two different fan modes: at maximum fan speed of 1230 RPM and in quiet mode at 800 RPM.
We have also included a very efficient tower cooler called Zalman CNPS10X Performa. Unlike Noctua NH-D14, it falls within the same price range as our today's newcomers.
Zalman CNPS10X Performa was tested with its default fan at 1100 RPM and at maximum speed of 2060 RPM. Besides that, all coolers except the boxed one from Intel were tested with two Thermalright TR-FDB fans at 1100 RPM and 2040 RPM.
The results of our comparative cooler testing are given on the diagram and in the table below:
Click to enlarge
At first let’s discuss the results demonstrated by our testing participants with their default fans. As we have expected, the least efficient cooler is the reference boxed cooler that comes with the Intel Core i7-980X with the fan working in the moderate speed mode (800-1800 RPM). However, we have to give this cooler due credit for coping with pretty seriously overclocked six-core processor at increased core voltage. If you switch the fan into Performance mode reaching the maximum speed of 2640 RPM, the peak CPU temperature will drop by another 6°C, which is quite good.
One step closer to the top is Scythe Yasya cooler. In the moderate acoustic mode of its default fan working in 500-1400 RPM speed interval, Yasya cools 1°C better than the reference cooler of our six-core CPU at its maximum speed. If we expand the fan rotation speed range to the maximum (720-1960 RPM), Scythe Yasya gains very little: the peak CPU temperature drops by only 3°C. As a result the new cooler from the Japanese maker takes only the fifth place out of six at this point.
Spire TherMax Eclipse II looks quite fine as it outperforms Scythe Yasya by 1°C in quiet mode with two fans at 1100 RPM and by 6°C at maximum fan speed. Of course, in this case the noise from the Spire cooler is higher than that from Scythe. Zalman CNPS10X Performa does even better: with one default fan it manages to successfully compete against TherMax Eclipse II with two fans onboard, even though the fan of the Zalman cooler rotates with lower maximum speed. Noctua NH-D14 with its default fans is again ahead of everyone else, besides the roaring Zalman and Spire coolers. And as for Baram-2010, it comes without a fan by default, so I guess it is time to move on to comparing the coolers’ performance when tested with two identical Thermalright fans.
Unfortunately, this part of our test session produced less interesting results than we have expected, because they turned out all very close. We can notice some differences in quiet fan mode at 1100 RPM. There are two “outsiders” here: Scythe Yasya and Spire TherMax Eclipse II, a “medium runner” – Noctua NH-D14 and two “leaders” – Zalman CNPS10X Performa and ThermoLab Baram-2010. I put all three words in quotation marks for a reason: the difference between these three coolers is less than 3°C, and at maximum rotation speed of our Thermalright fans all five coolers are almost equally efficient. So, what shall we do? The answer is fairly simple: we have to test our processor with each participating cooler to check out how far we will be able to overclock it, especially since its potential is far from being exhausted.
Well, Scythe Yasya proved capable of pushing the clock frequency of our six-core CPU to 4488 MHz at 1.475 V core voltage. With two fans working at 1100 RPM (quiet mode), the temperature of the hottest CPU core hit the dangerous level of 92°C, and at 2040 RPM it dropped down to 82°C:
As we see, despite some statements made by several other reviewers, Scythe coolers are not just fit for quad-core CPUs, but can ensure stable system operation even with seriously overclocked six-core processors.
Spire TherMax Eclipse II cooler with heatpipe direct touch technology didn’t blow us away by its performance with an overclocked six-core processor. With two fans working at 1100 RPM (quiet mode), Intel Core i7-980X remained stable at 4440 MHz frequency and 1.45 V voltage, while its peak temperatures stayed at 86°C. At maximum fan speed the frequency increased only to 4488 MHz at 1.475 V Vcore and 84°C CPU temperature:
Zalman CNPS10X Performa proved more efficient than Spire and Scythe products. With two fans working at 1100 RPM (quiet mode), we managed to overclock our processor to 4461 MHz at 1.475 V Vcore and maximum temperature of the hottest core around 88°C. At maximum fan rotation speed the CPU frequency hit the today’s record level of 4512 MHz at 1.475 V core voltage and 82°C temperature:
No matter how surprising it may seem, Noctua NH-D14 didn’t perform best of all today having taken the proud second place. With two fans working at 1100 RPM (quiet mode), Noctua NH-D14 and ThermoLab Baram-2010 demonstrated absolutely identical results during maximum CPU overclocking test: 4488 MHz frequency at 1.475 V Vcore and 87°C processor temperature. However, at the maximum fan speed of 2040 RPM Baram-2010 managed to outperform its neck-and-neck rival by 2°C:
What a surprise! I am sure no one expected anything like that to happen. The modest and relatively lightweight Baram-2010 outperformed the monstrous NH-D14! At the same time, it is important to understand that these results are obtained on a CPU with an uneven heat-spreader, so the efficiency of cooling using any of the tested products depends significantly on the quality of contact between the cooler base and the processor heat-spreader. Of course, ideally flat base of Baram-2010 has a larger contact spot on the Intel Core i7-980X heat-spreader than the slightly protuberant base of the Noctua NH-D14 cooler (which you can clearly see from the imprint photos above). This fact certainly has its say in the end result of the test.
Therefore, in one of our upcoming articles we are going to discuss the results obtained on a CPU with a perfectly lapped heat-spreader and how they compare against the results obtained today. Hopefully, a CPU with a lapped heat-spreader will overclock even better. And in the meanwhile let’s talk a bit about the acoustic performance of the today’s testing participants.
Due to smaller distance between the noise-meter and the fan rotor of 150 mm, we can now measure the noise from the coolers when tested not only with one, but also with two fans (the second fan being installed on the back of the heatsink). The graph below shows the obtained results:
If we put low-speed Noctua cooling fans aside, we can conclude that Spire cooler has the next quietest fans onboard. It is a real pity that there are no devices for adjusting their rotation speed and that they do not support PWM controlling. Far not every user out there is ready to put up with two fans spinning at more than 2300 RPM in their home system. Zalman and Scythe fans generate about the same noise, but Zalman seems to sound more pleasant. The loudest cooler of all tested today would be the boxed Intel cooler that comes with the Core i7-980X Extreme Edition processor. At high rotation speed its fan starts clinking, so that only the “Q” mode (800-1800 RPM) is in fact acceptable. And as for the noise from the 140 mm Noctua fans, we have already discussed it before: they remain pretty quiet up to 820-840 RPM and start producing too much noise at higher speeds.
In conclusion to our today’s review I would like to sum everything up about each of the new cooling solutions discussed and tested today. The Japanese Scythe Yasya comes with a not very good retention for Intel platforms. It is a pity that after Mugen 2 the company has gone back to weak plastic push-pins that bend the mainboard PCB. Moreover, among other drawbacks we could also mention the absence of grooves for the heatpipes in the copper cooler base, the absence of two additional wire clips for the second fan and missing shock-absorbing strips among the bundled accessories. Other than that it is a pretty decent cooler, very efficient, equipped with a fan supporting convenient rotation speed control and adjustment option. The price of the new Scythe Yasya is also absolutely justified.
Spire TherMax Eclipse II is one of the existing modifications of cooling solutions using heatpipe direct touch technology. Although it is also no leader today, it doesn’t suffer from any obvious design issues (except for the intervals between the heatpipes in the base part of it). I would like to point out that it is the only cooler tested today that comes with two high-quality fans and is priced no higher than Scythe Yasya. The only thing missing in TherMax Eclipse II is the ability to adjust the rotation speed of its fans or at least PWM control support.
The only thing we can say about a boxed cooler that comes with the Intel Core i7-980X extreme Edition processor is that it can cope with this exact processor in its nominal mode and during moderate overclocking, although you will have to put up with some serious noise in the latter case. I have t remind you that this is not a universal cooler and it is very unlikely to ever be selling separately from the Extreme Edition CPUs.
Finally, the cooler that deserves our ultimate admiration – ThermoLab Baram-2010. Its effective heatsink surface has become larger, the copper heatpipes and copper base have better and more precise contact, its retention mechanism has become simpler and even more universal than before (there was no LGA1156 support in the previous cooler models). While the cooler weighs only 710 g, it manages to cool the overclocked six-core processor just as good as the long-term leader – Noctua NH-D14, and at maximum fan rotation speed, it even outperforms the Noctua product. True, Baram-2010 costs quite a lot for a fanless product, but being one of the best (if not absolutely THE BEST) it is totally fair to ask $50 for a product like that.
Therefore, we are proud to award ThermoLab Baram-2010 with our Editor’s Choice title as one the best air coolers processors: