by Sergey Lepilov
04/15/2009 | 05:22 PM
The newest CPU cooler from the Japanese Scythe Co., Ltd. – Kabuto was first showcased at CeBIT 2009 and is yet another cooler with M.A.P.S. technology. This technology was first introduced in a well-known and very efficient Scythe Mugen 2 cooler. So, today we are going to talk about the new Kabuto. “Kabuto” in Japanese means “samurai helmet” that used to look like this:
There is even a Japanese saying: Katte kabuto no o o shimeyo ("Tighten the string of the kabuto after winning the war") that is interpreted as “don't lower your efforts after succeeding” and could be compared to “not to rest on one's laurels”. As we know, one can rest on laurels only after winning the victory, and our today’s article is going to reveal if Kabuto will win it or not.
The new cooler is shipped in a small cardboard box with mostly red colors:
The first thing that caught my eye on the cooler package was the blue “Andy” logo. As I learned, it was first designed when they created Andy Samurai Master cooler and “Andy” is the short for “Andreas”, from the name of one of the company employees – Andreas Bunen. This is a very creative way of rewarding the contribution to the company’s well-being. Moreover, Scythe marketing team saw it as a great opportunity. Europeans are a rare sight for Japan that is why they have also placed a photo of their colleague on the box (these samples were only distributed in Japan). Now Andy is just as good a brand as Scythe, although it has its own independent life.
As usual, Scythe covered the box with all sorts of info about the new cooler, including the detailed technical specifications, description of employed technologies and list of compatible platforms:
By the way, the photo on the right with the airflow schematics seems to have the colors mixed up.
Besides the cooler, there are the following accessories inside the box:
Scythe Kabuto is made in Taiwan and its recommended price is set at $48.
I have to say right away that Scythe Kabuto doesn’t look anything like a samurai helmet, so don’t try to wear it as one: it will not hold (only if you use some flexible ties or sticky tape) and will not protect you against any blow to the head. However, it looks very much like other top-coolers, i.e. the solutions designed with the airflow aimed at the mainboard PCB:
The cooler measures 124 x 133 x 132 mm and weighs 730g. It uses 6 copper heatpipes, each 6mm in diameter, that come out of one side of the cooler base.
The heatpipes hold three heatsink arrays, each made of 53 plates. The distance between two closest plates is 2mm and each plate is 0.35mm thick. The splitting of the cooler heatsink into several individual sections is called M.A.P.S. (Multiple Pass-Through Airflow Structure) and is supposed to lower the airflow resistance. It should also ensure fast dissipation of heat from the area around the heatpipes, which improves the efficiency of each heatsink array and the entire heatsink in general.
Our tests showed that this technology is extremely efficient in Scythe Mugen 2 with the heatsink built following the same principle. Moreover, it is evident that the production cost is way lower when the heatsink is built out of small plates rather than large plates, like by the competitor solutions (this is, probably, why the recommended pricing of both, Scythe Mugen 2 and Kabuto, is so low). Also, note that the heatpipes pierce the heatsink arrays on two levels, which allows to better distribute the heat more evenly over the heatsink.
The calculated effective heatsink surface (including the lower aluminum heatsink measuring 50 x 76 x 43 mm) is 7,057sq.mm. It is very good even for a tower-cooler, not to mention the top-designs, of course.
To ensure sturdiness of the whole structure, the heatsink arrays are connected with each other by means of small pieces of 12 plates. The frustrating thing is that the side plates are not sitting tight on the heatpipes and threaten to come off them. Too bad that the plates are simply pressed against the heatpipes, and not soldered to them.
The base of the helmet-cooler has no anticipated grooves for the heatpipes. They used thermal glue for them instead of the soldering technique:
The heatsink is topped with a 120-mm fan. It cools not only the entire thing, but also lowers the temperature of the components around the CPU socket, unlike the tower-coolers:
The nickel-plated cooler base is covered with special film protecting the base surface from scratches other physical damage. And actually, there is a lot to protect...
The evenness of the base plate and its finish quality are on unattainably high level for most other makers:
The imprint of the thermal interface applied to the processor heat-spreader looks as follows:
Scythe Kabuto is equipped with exactly the same fan as Mugen 2 – a nine-blade Scythe Slip Stream 120 with PWM rotation speed control (SY1225SL12LM-P model).
The fan is built with a slide bearing with 30,000 hours or more than 3.4 years MTBF of non-stop operation.
The fan rotation speed varies automatically in the interval between 200 and 1300 RPM at 74.25CFM airflow. The noise level from the fan shouldn’t exceed 26.5dBA.
The fan is fastened very securely with two wire clips inserted into special holes in the heatsink.
The ends of these clips go into the internal fan holes, so if you decide to replace the default Slip Stream 120 with something else and don’t want to invent an alternative retention for it, make sure that the fan you pick doesn’t have spindles instead of holes (SilverStone FM121, Scythe Gentle Typhoon).
This is pretty much all we can tell you about the cooler. Let’s move on to the installation techniques.
Scythe Kabuto can be installed not only on all contemporary platforms, but also on pretty outdated Socket 478 mainboards. You won’t need to remove the mainboard from the system case in any of the cases, because it uses the retention kits of the following types:
LGA775 and LGA1366 mainboards use the same retention. However, since Intel Core i7 mainboards have the cooler retention holes spaced out at a great distance, you will need to shift the plastic spindles with locks farther in the retention slits:
Unfortunately, Scythe Kabuto doesn’t have the same reliable retention with superb hold as Mugen 2 (with screws tightened to the backplate through the PCB). At the same time, the distance between the holes in the cooler base that are used to hold the retention bracket measures 45mm, which is exactly the same as the distance between the retention holes on Mugen 2. Therefore, you can easily use the retention from “second infinity” if you have it at your disposal.
Scythe recommends installing Kabuto into tower cases with the heatpipe ends facing up:
They promise maximum cooling efficiency in this particular position. The cooler may also be installed with the heatpipes positioned horizontally, but not with the heatpipes facing down.
This is what Scythe Kabuto looks like inside the system case:
The results of our tests showed that with the cooler installed in this particular way the peak CPU temperature was 3-4°C lower than with the cooler heatpipes turned horizontally:
The distance from the lower heatsink plate to the PCB surface is 60mm. The additional heatsink at the bottom doesn’t get beyond the main heatsink array, that is why Scythe Kabuto won’t interfere with any PCB components around the processor socket.
We tested the new cooling system from Scythe and its competitor inside a closed system case. 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 or open testbed 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 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.
Now let me say a few words about the today’s main competitors for Scythe Kabuto. First, we decided to include the Thermalright SI-128 SE top-cooler that didn’t participate in the recent super-cooler shoot-out, but definitely deserves being treated as such. Since SI-128 SE is incompatible with LGA1366 mainboards, we used the retention from IFX-14. Second, we also included the results for Thermalright IFX-14, which is the most efficient cooler today and will be used as a reference point for the sake of comparison.
Thermalright SI-128 SE and IFX-14 coolers were equipped with one and two SilenX iXtrema Pro IXP-76-14/16 fans 120 x 120 x 38mm in size. They were working at 930RPM and 1410RPM. We tested Scythe Kabuto with exact same fan, too. Thermalright SI-128 SE was also tested with the PWM fan from Scythe Kabuto cooler.
You may wonder what about the Mugen 2? Unfortunately, we couldn’t use the cooler with the best efficiency-to-price ratio as a reference this time. The thing is that installing Mugen 2 onto an LGA1366 mainboard requires removing the standard backplate and processor retention plate, which in its turn calls for taking the CPU out of its socket. Since we test a lot of coolers in our labs, we often remove the CPU from the socket. Of course, it has certain negative effect on the life span of the socket contact pins (I have already bent one of them, but fixed it). A regular user who replaces the cooler once every year and a half or even more than that, can easily install and remove Mugen 2 without threatening to damage the mainboard PCB. As for me, I can’t do that too often. Moreover, the backplate from the MUgen 2 retention kit is not designed for multiple installations, because its retention loops get pressed against the mainboard PCB which lowers the pressure on the cooler. Therefore, you may need to put additional washers beneath them eventually. I believe the reasons described above are sufficient to justify the replacement of our reference cooler with Thermalright IFX-14.
Before we move on to the main part of our test session, I would like to say the following. The thing is that we received the first Scythe Kabuto sample three weeks ago, which means that you may have been able to read this review way earlier. However, the cooler was defective, namely, demonstrated pretty average cooling efficiency. We checked it inside out, installed in different positions, used an alternative “through-retention” from Mugen 2 and different types of fans, but all our efforts were vain: we got the following graph for the beginning of CPU burn mode from the monitoring tools:
In other words, after only three Linpack runs the CPU temperature increased to 96°C and we had to terminate the test. We could feel that all cooler heatpipes warmed up much slower and cannot transfer the heat effectively to the heatsink plates. Our first Kabuto sample most likely performed so poorly because of the improper contact between the cooler base and the heatpipes. Nevertheless, even this defective sample equipped with a 1320RPM fan could cool our Core i7-920 processor overclocked to 3800MHz with the peak temperature around 89°C. The second Scythe Kabuto sample we managed to get our hands on proved 14°C more efficient than the first one right from the start. I hope that only one of the pre-production samples suffered from this issue and no defective products will get into retail.
Here we are going to check how the cooling efficiency of the new Scythe Kabuto cooler depends on the type and rotation speed of the cooling fan. To accomplish this, we tested the cooler in passive mode (no fan), with two 120 x 120 x 25mm Scythe Slip Stream 120 fans and with one 120 x 120 x 38mm SiloenX iXtrema Pro IXP-76-16 fan:
We ran the tests inside the system case with the side panel removed in order not to affect the fans efficiency in any way. Here is the results chart:
As you see, Scythe Kabuto can cope quite well with a Core i7-920 processor in its nominal mode even without a fan, while adding a 120-mm fan at practically noiseless 540RPM lowers the peak CPU temperature by more than 24°C! Moreover, we could overclock our CPU by 42.4% to 3800MHz in the same fan mode. Speeding the fan up to 860RPM and then to 1320RPM lowered the CPU temperature to 81°C and 75°C respectively, and allowed even greater overclocking. The maximum CPU frequency we could hit was 3970MHz at 1.35V Vcore, which I personally consider a very good result. I would also like to say that by replacing the standard 120 x 120 x 25mm fan with a bigger 38mm fan doesn’t have the desired effect even despite the higher rotation speed of the latter (1760RPM vs. 1320RPM). The latter indicates that the cooler heatsink is perfectly optimized for the Slip Stream fans.
Here I would also like to comment on the way the efficiency of a top-cooler depends on the presence or absence of the system case side panel. Tower-coolers which airflow is directed along the mainboard PCB are not as sensitive to this factor (we are talking about system cases with good airflow organization). The top-coolers are way more sensitive to that. Look for yourselves:
It turned out that Scythe Kabuto cools the CPU 4°C better under maximum load when the side panel is removed. Tower-coolers demonstrate half the difference of only 2°C. However, everything we have just said is true for our test system case – Antec Twelve Hundred. The results may vary by system case.
Now that they released the new BIOS version for the board, we can lock the processor clock frequency multiplier at 21. As a result, we could achieve higher CPU speeds at lower voltage settings than we did before. During the Linpack test we managed to overclock our 45 nm quad-core processor to 3.9 GHz (+46.2%). The nominal processor Vcore was increased to ~1.33125 V in the mainboard BIOS (+10.9%):
During CPU overclocking we activated in the mainboard BIOS the “Load-Line Calibration” function that lowers the voltage drop on the part of the voltage regulator circuitry before the CPU. The system memory voltage was locked at 1.55V and its frequency was at 1481MHz (8-8-8-18 timings). All other parameters available in the mainboard BIOS and connected with CPU or memory overclocking remained unchanged (set to Auto).
Let’s see what results we got here (the side panel of the system case has been removed for the time of tests):
Well, frankly speaking, Scythe Kabuto is no conjuror, but it performs just as well as one of the best top-coolers – Thermalright SI-128 SE. the latter gets a little bit ahead at high fan rotation speed because of higher heatsink plate density that benefit from higher air pressure. Here I would also like to stress how good the fan choice is for Scythe Kabuto cooler: installing a more powerful SilenX iXtrema Pro instead of the default Slip Stream doesn’t improve the cooling efficiency that much at all. As for the reference cooler, Thermalright IFX-14, it easily leaves its competitors behind in all test modes. However, we have expected something like that. It will win even more inside a completely closed system case.
The diagram below shows the acoustic readings for our today’s testing participants:
Since Scythe Kabuto fan reaches maximum 1320RPM only under heavy CPU utilization and in regular mode works at about 300RPM, we can consider the new cooling solution a quiet system. Of course, if your CPU is constantly working under heavy load of resource-hungry applications, the fan will rotate at a speed close to maximum and you will hear it against the background of a quiet system case.
Of course, none of the real Japanese samurai warriors could ever imagine what would be called “Kabuto” these days. However, overclocking fans shouldn’t worry about it. It is Scythe’s competitors who should be concerned, because the new Scythe Kabuto cooler we tested today proved extremely efficient and pretty quiet at the same time. If we compare Kabuto with SI-128 SE, the former will offer the same efficiency at a much lower price, comes with a fan, is compatible with all existing platforms, boasts ideally even mirror-shining base surface and is installed in a much easier manner. As for the clip-retention for LGA1366 platforms, we cannot complain about bending of the PCB anymore, because there is a default backplate on all LGA1366 boards (that is why we didn’t benefit much from using the Mugen 2 screw-retention). However, it still remains an issue for LGA775 platforms.
The only drawback that we can find about Scythe Kabuto is the absence of grooves in the base plate that could ensure better contact with the heatpipes. However, there is one more thing we would like to say at this point. What if they designed a tower-cooler using the Kabuto heatsink concept? Just take two Kabuto heatsinks and install a 120-mm fan between them. The length and width of this cooler will be 120 and 129 mm respectively, but the effective surface of the new “Kabuto Tower” solution will double and the fan will work way more effectively than on Mugen 2. Moreover, I don’t think they will have to upgrade their production lines in order to manufacture the new cooler, because all they will need to do is take two Kabuto heatsinks, throw six heatpipes into the base plate and insert a fan between two parts of the heatsink. I really hope that Scythe engineers will at least consider this suggestion, because theoretically, a new cooler like that seems very interesting.
P.S.: As we found out from a small comparative fan testing, SilenX iXtrema Pro fans do not suit too well for the reference IFX-14, especially at low rotation speeds. They lose about 3-4°C under maximum CPU load to the Enermax Magma fans at 930RPM that we used before. Things get a little better at maximum fan rotation speeds, but again not in SilenX iXtrema Pro’s favor. Anyway, stay tuned for a big fan shoot-out we are preparing for you next, and trust me there will be a lot to choose from there.