ForewordIn my humble opinion, the Japanese Scythe Co., Ltd. is a leading manufacturer of air coolers for CPUs. To come to this conclusion, it is necessary to evaluate its main competitors. For example, Thermalright produces the most efficient coolers, but offers them at high prices, while not bothering to control the evenness of the bases, and has an underdeveloped dealer network, which is why it is often simply impossible to purchase its products, especially far from large cities. The well-known Korean company Zalman in the field of air cooling systems, by and large, has only a big name, deserved at the very beginning of the millennium. Thermaltake releases good coolers, but they do it quite rarely, although this situation has begun to improve lately. ZEROtherm and the new ThermoLab are all too rare on the market. cooler master, perhaps, is Scythe's most formidable competitor today, as its product range includes both excellent coolers in terms of price / performance ratio (Hyper TX 2 and Hyper 212), as well as expensive V8 and V10 supercoolers. In addition, two more new items will appear very soon, and the products of this brand are widely distributed around the world. Who else did you forget? Titan, ASUSTek, Noctua and Xigmatek - these companies also rarely spoil us with new products, and their products are not widely used on the market, with the exception of Xigmatek, which only produces coolers with direct contact technology that does not work well with all modern processors.
Unlike competitors, Scythe products can be purchased almost all over the world, and, compared to other brands, Scythe coolers stand out quite reasonable prices: cost of her coolers ranges from one to two thousand rubles, which is relatively small for products of this class (for comparison, more than half of the Thermalright coolers available in our store are more than two thousand rubles). The range of products is quite wide, from the neat Katana II and the ultra-compact Shuriken to the gigantic and very expensive Orochi. Updating the lines of cooling systems takes place with enviable constancy for other manufacturers. Every now and then Scythe announces this or that cooler. Of the new products already released, but not yet tested by us, we can note the Katana III (SCKTN-3000), REEVEN (RCCT-0901SP) or KILLER WHALE coolers. In addition, the company's product range includes a wide selection of fans of various sizes and purposes, as well as other useful accessories. Only one thing is missing - a cooler, which could be called the absolute leader among air cooling systems. But, as it turned out, with the release of Mugen 2, Scythe successfully closed this gap.
The first version of “infinity” (namely, this is how the name of the cooler is translated from English “Infinity”) appeared in 2006, far from the standards of the Hi-Tech industry. At that time, the Scythe Infinity cooler was generally recognized as one of the best in terms of cooling efficiency, if not the best. Almost a year later, the second revision of Infinity was released to the market, renamed “Mugen” - this word also means “infinity”, only now translated from Japanese. Then the changes affected only the fan (a more productive and lighter Slip Stream model was installed). Finally, at the very beginning of 2009, Scythe released the second version of the Mugen cooler, with a fundamentally new heatsink, a new fan and a different mounting system.
But first things first.
Scythe Mugen 2 cooler review (SCMG-2000)
Packaging and equipmentThe new cooler is sealed in a compact cardboard box with a picture of the cooling system on the front side:
Scythe Mugen 2 is depicted hovering in outer space against the background of the Earth, apparently personifying that same infinity. The other sides of the box are decorated in the same style, on which a description is given. key features cooler, specifications, as well as listing the accessories included in the package:
Among the latter, one can note a universal plate, sets of fasteners and screws, SilMORE thermal paste, two wire brackets for the fan and instructions for installing the cooler in six languages, including Russian:
Inside the package, all components are securely fixed, and there are cardboard inserts between the sections of the radiator, which minimizes the risk of damage to the device during its transportation.
The Scythe Mugen 2 is made in Taiwan and has a MSRP of just $39.5. In Moscow, at the time of preparation of the article, the cooler was not yet on sale.
Design features
The new cooling system belongs to tower-type coolers and has dimensions of 130x100x158 mm and weighs 870 grams together with the fan. The radiator looks like this:
It consists of five independent sections, each of which has one heat pipe with a diameter of 6 mm. Thus, there are five tubes in total. The distance between all sections of the radiator is the same and is 2.8 mm:
Actually, the division of one solid radiator into five separate sections is the key feature of Scythe Mugen 2. Japanese engineers called this feature M.A.P.S. (“Multiple Airflow Pass-through Structure”), which loosely translates as “a structure for passing multiple airflows”. According to Scythe engineers, such a “dissected” heatsink will facilitate not only the rapid outflow of heat from the heatsink areas adjacent to the tubes, but also reduce airflow resistance and increase the efficiency of each individual heatsink and the cooler as a whole. Separately, it is indicated that such a structure is the best fit for Scythe fans of the Slip Stream 120 series, one of which comes with Mugen 2.
Each radiator consists of 46 aluminum plates 0.35 mm thick with a spacing of 2.0 mm between fins:
The width of the three central sections is less than the width of the two extreme ones: 22 mm and 25.5 mm, respectively:
But the length of the radiator plates is the same and is 100 mm. Thus, the heat sink area of Scythe Mugen 2 is about 10.5 thousand square centimeters, which is noticeably larger than even that of the giant Scythe Orochi (about 8700 cm²), and comparable to the three-radiator Cooler Master V10 (also about 10,500 cm²).
I will add that the ends of the heat pipes are covered with figured aluminum caps.
At the bottom of the cooler, there is an additional aluminum radiator with dimensions of 80x40 mm, adjacent to the upper part of the tubes above the base:
Apparently, it is designed to remove the heat load from the surface of the tubes, which is located above the base and is not cooled by anything.
The tubes are glued to the base with hot-melt glue - we will probably never get the desired grooves from Scythe (by the way, there are grooves in the additional radiator). But the processing quality of the nickel-plated copper plate is at the highest level:
The surface of the plate is even, except that in the corners, when checking the evenness with a ruler, you can see scanty gaps:
Most importantly, there are no irregularities in the contact area between the base and the heat spreader of the processor:
Scythe Mugen 2 is equipped with a nine-blade 120x120x25 mm fan of the Slip Stream 120 series, model SY1225SL12LM-P:
The fan is based on a sleeve bearing with a standard service life of 30,000 hours (more than 3 years of continuous operation). The fan speed is controlled by pulse-width modulation (PWM) in the range from 0 to 1300 rpm, while the air flow can reach 74.25 CFM. The maximum fan noise level is declared at around 26.5 dBA.
Slip Stream 120 is attached to the heatsink using two wire brackets, the ends of which are inserted into the outer holes of the fan frame, and the brackets themselves snap into special grooves in the heatsink:
Moreover, in total, the cooler's heatsink has eight symmetrically arranged slots, which will allow you to hang four fans on the heatsink at once:
True, for this you will need 3 more fans and three additional sets of mounts.
As you understand, one complete fan can be installed either along the sections or across:
The maximum cooling efficiency will be achieved when the air flow is directed along the sections. It is this position of the fan that is recommended by the manufacturer, so the second option is possible only in exceptional cases, when for some reason it is impossible to hook the fan on one of the wide sides of the cooler.
Platform support and installation on motherboards
Scythe Mugen 2 can be installed on all modern platforms without exception, and even on an already outdated platform with a Socket 478 connector. Detailed instructions will tell you about the cooler installation procedure, but here we will consider its main points.
First of all, to install the cooler, you will need to screw fasteners to its base that correspond to the processor socket of your motherboard:
Socket 478Socket 754/939/940/AM2(+)/AM3LGA 775/1366
Further, schematically, the procedure for installing Scythe Mugen 2 on each of the platforms looks like this:
Socket 478LGA 775LGA 1366
Socket 754/939/940socket AM2(+)/AM3
As you can see, in all cases, the new cooler is attached to the plate on the back of the motherboard, so the latter will have to be removed from the system unit case. Scythe has finally ditched the flimsy and bending "push-pin" mountings on the motherboard and equipped its flagship with excellent mountings and a universal plate:
Despite the apparent bulkiness, it fit without any problems on the back side of the DFI LANPARTY DK X48-T2RS motherboard:
By the way, if the cooler is installed on motherboards with an LGA 1366 connector, the standard pressure plate of these boards will need to be removed and replaced with a plate from the Mugen 2 kit. A special key is supplied with the cooler to dismantle the standard plate.
The distance from the base surface of the cooler to the bottom plate of the heat sink is 41 mm, and the cooler is compact in the area of the base, so neither heat pipes nor an additional heat sink interfered with the installation of the cooling system on the board:
But there were problems when installing the fan on the radiator. Firstly, I had to remove the RAM module from the first slot, since its high heatsink did not allow hanging a fan, and secondly, one wire bracket at the bottom could not be hooked onto the heatsink, because it rested on the motherboard chipset heatsink:
However, the last problem is hardly serious - after all, the upper edge of the wire went into the groove. As for the memory module, I would recommend that potential owners of Mugen 2 either purchase modules without heatsinks, or make sure in advance that the cooler with a fan and boards with high memory modules are compatible. To help the latter, I'll add that the distance from the central axis of the cooler to the edge of the wide heatsink is 50 mm (and another 25 mm must be added to the fan).
Inside the case of the Scythe Mugen 2 system unit looks like this:
No fan lights and other tinsel for you. Everything is serious.
Specifications
The technical characteristics of the new cooler are summarized in the following table:
Test configuration, tools and testing methodology
The effectiveness of the new cooling system and its competitor was tested inside the case of the system unit. Testing was not carried out on an open bench and will not be carried out in the future, since in comparison with the temperatures inside the new case at low fan speeds, there was no difference with temperatures on an open bench at all, and at high speeds an open bench won back only 1-2 ° C, for which it certainly makes no sense to regularly sort through the system.
The configuration of the system unit during testing was not subject to any changes and consisted of the following components:
Motherboard: DFI LANPARTY DK X48-T2RS (Intel X48, LGA 775, BIOS 03.10.2008);
CPU: Intel Core 2 Extreme QX9650, (3.0 GHz, 1.15 V, L2 2 x 6 MB, FSB 333 MHz x 4, Yorkfield, C0);
Thermal interface: Arctic Silver 5;
DDR2 RAM:
1 x 1024MB Corsair Dominator TWIN2X2048-9136C5D (1142MHz, 5-5-5-18, 2.1V);
2 x 1024MB CSX DIABLO CSXO-XAC-1200-2GB-KIT (1200MHz, 5-5-5-16, 2.4V);
Video card: ZOTAC GeForce GTX 260 AMP2! Edition 896 MB, 650/1400/2100 MHz (1030 rpm);
Disk Subsystem: Western Digital VelociRaptor (SATA-II, 300 GB, 10,000 rpm, 16 MB buffer, NCQ);
HDD cooling and soundproofing system: Scythe Quiet Drive for 3.5" HDD;
Optical drive: Samsung SH-S183L;
Case: Antec Twelve Hundred (replaced stock 120mm fans with four 800rpm Scythe Slip Stream fans, 800rpm 120mm Scythe Gentle Typhoon at the bottom of the front wall, standard 400rpm 200mm fan on top );
Control and monitoring panel: Zalman ZM-MFC2;
Power supply: Zalman ZM1000-HP 1000W, 140mm fan;
All tests were performed under Windows Vista Ultimate Edition x86 SP1 operating system. Software, used during testing, is the following:
Real Temp 3.0 - for monitoring the temperature of the processor cores;
RightMark CPU Clock Utility 2.35.0 - to control the operation of the processor's thermal protection (clock skipping mode);
Linpack 32-bit in LinX 0.5.7 shell - for CPU load (double test cycle with 20 Linpack passes in each cycle with 1600 MB of RAM used);
RivaTuner 2.23 - for visual control of temperature changes (with RTCore plugin).
So the full screen shot during testing is as follows:
The processor temperature stabilization period between test cycles was approximately 10 minutes. The maximum temperature of the hottest of the four cores of the central processor was taken as the final result.
The room temperature was controlled by an electronic thermometer installed next to the housing with a measurement accuracy of 0.1 °C and the ability to monitor changes in room temperature over the past 6 hours. During testing, room temperature fluctuated in the range of 23.5-24.0 °C.
A few words about the cooler with which we will compare Scythe Mugen 2. It is said that the heat pipes of this cooler are filled with gas delivered from one of the satellites of Jupiter, and that one of the Formula 1 teams decided to use it in the 2009 season to cool the KERS system. .. All we know for sure is that its name is ThermoLab BARAM, and so far it has been the best cooler among those that have been in our hands:
BARAM was tested with one and two Scythe Slip Stream 120 fans at speeds from 510 to 1860 rpm. Scythe Mugen 2 was tested with the same fans and in the same speed modes, in addition to tests with a standard PWM fan.
Cooler Efficiency Test Results
When tested with Linpack, the overclocking limit of a 45 nm quad-core processor at a minimum fan speed of 510 rpm coolers was 3.8 GHz (+ 26.7%) with an increase in the motherboard BIOS voltage to 1.5 V (+30 ,four %):
None of the two coolers tested today could cope with one very quiet 510 rpm fan to cool the overclocked processor, so the results “start” from the mode of operation of coolers with two such fans:
That's it! More recently, ThermoLab BARAM outperformed the Thermalright Ultra-120 eXtreme, even if only slightly, and today Scythe Mugen 2 outperformed BARAM by 2°C. Another change in the leader and standard among air-cooled systems. Pay attention to how well the fan is chosen for the new cooler. With two 860 rpm fans, Mugen 2 cools the processor 2 °C worse than with a single PWM fan with a maximum speed of 1300 rpm. Installing an even more powerful 1860 rpm fan results in a 3°C drop in temperature, but the noise level becomes quite high. Well, the second powerful fan does nothing at all in terms of cooling efficiency.
The “second infinity” turned out to be more effective than the “air flow” when testing for maximum processor overclocking:
Scythe Mugen 2 (2х1860 RPM)ThermoLab BARAM (2x1860RPM)
If in the future we will witness such frequent changes in the leaders of air cooling systems, “pinching off” a couple of degrees Celsius each time, then over time the coolers will reach unprecedented heights in the field of processor cooling.
Conclusion
When preparing conclusions for articles on testing cooling systems, I always try to start by listing the cooler's shortcomings, and only then talk about their advantages, but today it turned out to be very difficult to find shortcomings in the reviewed and tested Scythe Mugen 2. You can find fault with the lack of another pair of wire brackets in the kit for installing a second fan, or with the cheap and not very efficient SilMORE thermal paste, or with the lack of grooves for tubes in the base of the cooler... However, all these shortcomings pale in front of the cooler's unsurpassed efficiency, low noise level at maximum load on the processor and noiselessness at normal work, really low cost compared to other supercoolers, full compatibility with all platforms and, finally, the wide distribution of Scythe products around the world. If you try Scythe Mugen 2 against ThermoLab BARAM in all these parameters, then it is obvious that the (now former) standard loses in all respects. However, I still propose to draw the final conclusions after a large-scale testing of the top ten supercoolers on a platform with an Intel Core i7 processor, which will soon await you.Check the availability and cost of Scythe coolers
Other materials on this topic
Review of Coolers Thermaltake TMG IA1 and Scythe Kama Angle
Thermalright AXP-140: High Efficiency Low Profile Cooler
Cooler Master V10: 10 heat pipes, 3 heatsinks, 2 fans and a Peltier module. Supercooler?
Finding the optimal places for placing fans in a given case.
I tried for myself. So that the data does not disappear, I designed it in an article.
Pictures fictitious from the Internet (there are no pictures of my own).
I got the idea for the experiment from here.
Results table.
With a list of hardware, software and fan installation locations.(a little larger table is attached at the bottom of the page)
Text description
Case appearance
Cooler Noctua NH-D14
With one NF-P12, blowing through both towers. Thermal grease Zalman STG-2
Vertical CPU cooler options
Initially there were two fans.
Noctua NF-P12 and Cooler Master A12025 (hereinafter referred to as CM).
I put P12 on blowing from the back wall, and CM on blowing through the bottom.
Then I tried to pick up such a load that, with LinX + Kombustor, the system, if not sewn up, would noticeably overheat.
Bringing the CPU to 90C was not difficult.
Stable load 100%, 3.5GHz.
But the frequency of the core of the video card twitches when running LinX + Kombustor at the same time (Kombustor itself presses very calmly). Anyway. I added +100MHz to the GPU core in MSI Afterburner to warm up and get those 76.4C / 88.6C core / VRM at 1921 rpm of the video card coolers.
I took the LinX settings and the frequencies of the CPU, GPU in this variant as starting points (reference point), and did not change the parameters anymore. This option was tested up to 7 successful times in order to fill in the statistics and so far I figured out in what ranges the warmed-up system is playing. Sometimes the video adapter gave out some overexcited porn from its storerooms. I discarded such data, took the average from the rest, rounded to tenths. Therefore, in the table, values \u200b\u200bwith a comma.
The power supply has a bottom fence, an exhaust at the back. Works quietly. It did not consider it advisable to draw warm case air through it, so the PSU did not turn it over. I would like to know its temperature and speed, but there is nothing to approach, the monitoring programs do not take the data of this PSU, they do not show it :(
It was the hottest, indicative option (only with 2 fans). Further - cooler.
Another Noctua NF-P12 has appeared.
I put it in the classical way on blowing on the front (front) panel above, and CM below.
One of the hard drive walls has been removed.
And only the second non-removable wall with large oval holes prevented the flow of P12.
At the bottom, the SM entered into a head-on battle with the HDD and SSD. All of its 1200 rpm went into conquering the best HDD temperature for this variant. 
SM dropped the HDD and settled on the side wall (in the left mounting location). Its diameter is about a quarter blocked at the bottom of the PSU. It blows on the motherboard, which made it colder MB -5C, PCH -4C.
HDD took offense and warmed up by +2C.
The video card prefers to be silent.
- - - - - - - - - - - - - - - - - - - - - -
SM moved to the right mounting place along the wall of the case.
MB scored +4C, PCH also +0.8C
.
- - - - - - - - - - - - - - - - - - - - - -
The valve NF-P12 also moved to its side, to the left of the CM.
Together, from the sidewall, the guys blew much stronger than being in the corral of the labyrinths of the front panel.
So, in comparison with A-2/1-a: the mother has cooled down by -4.3C; PCH for all -10.8C;
even vidyaha with VRM said -2.7C and -2.3C.
Deprived of direct and curved airflow, the HDD freaked out at + 2.7C, but its antics at 31.3C are naturally sidelined by everyone.
By the way, he saw a quiet 5400rpm and 38 degrees maximum only in the meanest version with 2 valves.
Although he was not given frantic reading / writing tasks, there was no reason to warm up.
- - - - - - - - - - - - - - - - - - - - - -
The violent head knocked out crazy handles to stick 2 sheets of A4 from the bottom of the valves on the sidewall - right under the video slot, across its entire width. Say, so all the air pumped in by two 120-kami will be along the guide, without loss, support both regular turntables of the video card.
Mom threw off the degree. PCH dialed +7.4C apparently, a sheet of paper directed the flow past him.
HDD still inserted its + 1.7C.
Vidyakhino's achievement of -0.5C is not worth such "modding".
- - - - - - - - - - - - - - - - - - - - - -
I remembered that I managed to seal the top cover with adhesive tape (from dust). Like all the slots inside the case after purchase.
I removed the adhesive tape from the lid, there was a metal mesh with 2mm holes.
Helped. By convection through the lid. Warm air can be felt on the hand.
Finally, the CPU came into motion, though only -0.8C. Mom also dropped the degree. PCH at -6.8C eased.
- - - - - - - - - - - - - - - - - - - - - -
I separated the metal mesh from the lid. There was a frame with large holes in the form of honeycombs 21x23mm.
And all the components still unanimously dropped from -0.6 to -1.5 degrees.
So, in this version, the coldest indicators are CPU, MB, and GPU. And free breathing through the top makes sense.
.
- - - - - - - - - - - - - - - - - - - - - -
By the way, the CPU noticeably reacts only to shifts in the upper part of the case, and the video card - to rearrangements in
bottom half. The vidyahi brick just divides the body into 2 fronts, upper and lower.
Another crazy idea is to organize an air duct/shroud, through which the air flow through the CPU cooler will be isolated, without dissipating hot air on the towers.
Everyone immediately became ill. From +4.1C on CPU to +1.1GPU.
Horizontal CPU cooler options
Actually, a dream. Expand the towers to blow through the roof. I read that it would be okay.
Okay started cracking right away. So far, I have only deployed the cooler, and left the exhaust NF-P12 on the back wall.
Compare, for example, with the winning variant A-2/1-g(convection through honeycombs in the lid). Prots hung himself and scored +11.4C, the rest is insignificant. Unless VRM smiles. This is probably his tower valve -2.5 degrees sucked. This valve is just tight between the video card cover and the tower of its cooler - it suffocates, there is nothing to pump.
- - - - - - - - - - - - - - - - - - - - - -
NF-P12 from the back rushed to the roof, above the radiator towers - pulling a dream. Pull through
perforation 2mm. I don’t like the honeycomb holes on the lid, so I removed the mesh only for the test in one
option ( A-2/1-g). The perforation on the back wall (now without a valve) was sealed with adhesive tape.
Such a maneuver removed only -1.3С from the CPU, which is up to the light bulb. The video card with its VRM misunderstood something and added +1.3 and 2 degrees, respectively. Mum got a degree hotter. Okay, another trump card in your pocket.
- - - - - - - - - - - - - - - - - - - - - -
On the CPU cooler, remove the NF-P12 valve from the video card cover and put it inside, between the radiator towers.
From here it pumps much better.
Compared to the previous version: saves percent by -7.8C.
True, it stops sucking VRM, which scored its + 2C.
Results
With a given number of fans, the winning variant is A-2/1-g.And this is: 2x120 blowing through the side wall, 1x120 blowing out from behind.
The orientation of the CPU cooler is vertical (blowing onto the rear wall valve).
Gives the best results for CPU, MB, GPU temperatures.
At the same time, HDD, PCH and VRM temperatures are not far behind competitors.
Worst case A-1/1(with two blow-in-bottom/blowing-back fans).
Two turntables, of course, spit weakly. Moreover, Cooler Master (CM) with its breath at 1200rpm does not look menacing. Comparing it side by side with the Noctua NF-P12 on the side panel, covering the holes in the perforation with your hand, the CM is all the same, and the Noctua already whistled, greedily sucking in air. Working on blowing from the rear wall, the CM also did not excel, so in the tests it constantly pumped out NF-P12 there.
The temperature difference between the best and worst options in degrees:
CPU -12.6
MB-13.9
HDD-6.6
PCH-21.2
GPU -17.2
VRM-13.1
outdoor stand
A case without two side walls, a cover, and without all three case fans.I remembered it at the very end. Thought - skunk to my winning variant.
But it was not there.
As an option A-2/1-g"extinguishes" an open stand:
CPU +0.9
MB-5.8
HDD -3.8
PCH-11.5
GPU -3.8
VRM-2.5
It seems that components without active airflow do not feel so comfortable.
Only percent exhaled, almost 1 degree.
- - - - - - - - - - - - - - - - - - - - - - .
I'm not a special tester, and I switched to a system unit recently after 9 years on laptops.
Therefore, jambs and inappropriate conclusions can be enough. Be carefull.
Thank you for attention.
Closest forum topic
Bonus
Checking the two options Romulus.
A-1/2-a and A-1/2-b
We unfold the left valve on its side for blowing.
Hard case. Ran the test 4 times. It seems that the system depends on the wind, where it blows, such are the numbers. Usually, for 3 runs at different times, completely settled, almost identical values were obtained. And this…
I had to stick my face closer to what was happening.
It's such a bullshit. At the exit from the sidewall, the air is strongly sprayed like a fan on the sides. And next to the intake valve. And he steals some of the spent exhaust. Especially if there is a slight movement of air in the room, for example from a window, at least a little bit licks on the side of the body, and even from the exhaust to the retractor - intestinal volvulus is guaranteed. Unstable cooling.
GPU 64.3C is almost like an open bench, it was worse only in the version with 2 fans.
CPU 80 is slightly better than in the "leather".
Retractable from the side we throw to the bottom.
I did not seal the space freed from the fan on the side. But I checked. There is a small air leak through it. A thin check from the store does not hold, but it tries, it sticks slightly to the perforation.
Proc 80.3C Something he doesn’t like the injection crack at the bottom, neither in this version, nor in the previous one. It's hot under the roof, if you don't pump it in from below, or what?
Results, mails are identical to the previous option, within 1 degree.
- Inspector Petrenko. Your documents. Violating...
- Chito violating the nayalnika?
We're breaking the balance!
- Acid-alkaline?
- Not. Supply and exhaust!
All out. That is, both turntables on the sidewall are exhaust. The entire supply is unofficial, through the cracks.
Prots and mother pulled themselves up, the rest sank.
CPU 76C. -1.3C colder than the best result in the table. It seems that if the non-optimal "inversion of the intestines" at the bottom of the case is stupidly sucked out with two valves, then the percent will provide for itself.
MB threw off the degree and also set an intra-table record on this moment 40.3C The sensor under the hood sucked something.
HDD 35.8C warmed up ugly; RSN 47.1C
GPU 65.8C. Didn't stand out at all. Some kind of conflict of interest. 2 video card helicopters row themselves. And 2x120 is right next to it, on the sidewall - they are pumped out of the case. And what to eat vidyahe?
* * *
Total: alignment A-2/1-g remains in high esteem, although it slightly outperformed it in terms of CPU and MB A-0/3.
Will you be fourth?
Another NF-P12 has appeared.Took the option A-2/1-f(2 blowing in from the side, 1 blowing out from the back) and stuck this 4th valve to the bottom and front panel - blowing in, and blowing out on the lid.
The table shows that the effect is only when installed on the bottom. The GPU cooled down -2.5C, VPM -4.2C, and MB -1.4C.
Front injection or top exhaust with such a 4th fan - up to the bulb.
Often used to build a large radiator heat pipes(English: heat pipes) - hermetically sealed and specially arranged metal tubes (usually copper). They transfer heat very efficiently from one end to the other: thus, even the farthest fins of a large heatsink work effectively in cooling. So, for example, the popular cooler is arranged
To cool modern high-performance GPUs, the same methods are used: large radiators, copper core cooling systems or all-copper radiators, heat pipes to transfer heat to additional radiators:
Recommendations for choosing here are the same: use slow and large-sized fans, the largest possible heatsinks. So, for example, popular cooling systems for video cards and Zalman VF900 look like:
Usually, the fans of video card cooling systems only mixed the air inside the system unit, which is not very effective in terms of cooling the entire computer. Only recently, cooling systems began to be used to cool video cards, which carry hot air outside the case: the first steels and a similar design from the brand:
Similar cooling systems are installed on the most powerful modern video cards (nVidia GeForce 8800, ATI x1800XT and older). Such a design is often more justified, in terms of the proper organization of air flows inside the computer case, than traditional schemes. Air flow organization
Modern standards for the design of computer cases, among other things, regulate the way the cooling system is built. Starting with, the release of which was launched in 1997, a computer cooling technology is being introduced with a through air flow directed from the front wall of the case to the back (additionally, air for cooling is sucked in through the left wall):
Those interested in details are referred to latest versions ATX standard.
At least one fan is installed in the computer's power supply (many modern models have two fans, which can significantly reduce the rotation speed of each of them, and, therefore, the noise during operation). Additional fans can be installed anywhere inside the computer case to increase airflow. Be sure to follow the rule: on the front and left side walls, air is blown into the case, on the back wall, hot air is thrown out. You also need to make sure that the flow of hot air from the rear wall of the computer does not fall directly into the air intake on the left wall of the computer (this happens at certain positions of the system unit relative to the walls of the room and furniture). Which fans to install depends primarily on the availability of appropriate mounts in the walls of the case. Fan noise is mainly determined by fan speed (see section ), so slow (quiet) fan models are recommended. With equal installation dimensions and rotational speed, the fans on the rear wall of the case are subjectively noisier than the front ones: firstly, they are farther from the user, and secondly, there are almost transparent grilles at the back of the case, while various decorative elements are at the front. Often noise is created due to air flow around the elements of the front panel: if the amount of air flow transferred exceeds a certain limit, eddy turbulent flows form on the front panel of the computer case, which create a characteristic noise (it resembles the hiss of a vacuum cleaner, but much quieter).
Choosing a computer case
Almost the vast majority of computer cases on the market today comply with one of the versions of the ATX standard, including in terms of cooling. The cheapest cases are not equipped with either a power supply or additional devices. More expensive cases are equipped with fans to cool the case, less often - adapters for connecting fans in various ways; sometimes even a special controller equipped with thermal sensors, which allows you to smoothly adjust the rotation speed of one or more fans depending on the temperature of the main components (see for example). The power supply is not always included in the kit: many buyers prefer to choose a PSU on their own. Of the other options for additional equipment, it is worth noting the special fastenings of the side walls, hard drives, optical drives, expansion cards that allow you to assemble a computer without a screwdriver; dust filters that prevent dirt from entering the computer through the ventilation holes; various nozzles for directing air flows inside the case. Exploring the fan
Used to transport air in cooling systems fans(English: fan).
Fan device
The fan consists of a housing (usually in the form of a frame), an electric motor and an impeller mounted with bearings on the same axis as the motor:
The reliability of the fan depends on the type of bearings installed. Manufacturers claim the following typical MTBF (number of years based on 24/7 operation):
Taking into account the obsolescence of computer equipment (for home and office use it is 2-3 years), fans with ball bearings can be considered "eternal": their life is not less than the typical life of a computer. For more serious applications, where the computer must work around the clock for many years, it is worth choosing more reliable fans.
Many have come across old fans in which the plain bearings have worn out their life: the impeller shaft rattles and vibrates during operation, making a characteristic growling sound. In principle, such a bearing can be repaired by lubricating it with solid lubricant - but how many will agree to repair a fan that costs only a couple of dollars?
Fan characteristics
Fans vary in size and thickness: commonly found in computers are 40x40x10mm for cooling graphics cards and hard drive pockets, as well as 80x80x25, 92x92x25, 120x120x25mm for case cooling. Also, fans differ in the type and design of the installed electric motors: they consume different current and provide different impeller rotation speeds. The size of the fan and the speed of rotation of the impeller blades determine the performance: the generated static pressure and the maximum volume of air transferred.
The volume of air carried by a fan (flow rate) is measured in cubic meters per minute or cubic feet per minute (CFM). The performance of the fan, indicated in the characteristics, is measured at zero pressure: the fan operates in an open space. Inside the computer case, the fan blows into the system unit of a certain size, so it creates excess pressure in the serviced volume. Naturally, the volumetric efficiency will be approximately inversely proportional to the pressure generated. specific kind flow characteristics depends on the shape of the used impeller and other parameters specific model. For example, the corresponding graph for a fan is:
The simple conclusion from this follows: the more intensively the fans in the back of the computer case work, the more air can be pumped through the entire system, and the cooling will be more effective.
Fan noise level
The noise level created by the fan during operation depends on its various characteristics (more details about the reasons for its occurrence can be found in the article). It is easy to establish the relationship between performance and fan noise. On the website of a major manufacturer of popular cooling systems, we see that many fans of the same size are equipped with different electric motors that are designed for different rotation speeds. Since the same impeller is used, we obtain the data we are interested in: the characteristics of the same fan at different rotation speeds. We compile a table for the three most common sizes: thickness 25 mm, and.
Bold font indicates the most popular types of fans.
Having calculated the coefficient of proportionality of the air flow and the noise level to the speed, we see an almost complete match. To clear our conscience, we consider deviations from the average: less than 5%. Thus, we got three linear dependencies, 5 points each. Not God knows what statistics, but for linear dependence this is enough: we consider the hypothesis confirmed.
The volumetric efficiency of the fan is proportional to the number of revolutions of the impeller, the same is true for the noise level.
Using the obtained hypothesis, we can extrapolate the obtained results using the least squares method (LSM): in the table, these values are marked in italics. However, it must be remembered that the scope of this model is limited. The investigated dependence is linear in a certain range of rotation speeds; it is logical to assume that the linear nature of the dependence will remain in some neighborhood of this range; but at very high and very low speeds, the picture can change significantly.
Now consider the line of fans from another manufacturer:, and. Let's create a similar table:
Calculated data are marked in italics.
As mentioned above, at fan speeds that differ significantly from those studied, the linear model may be incorrect. The values obtained by extrapolation should be understood as a rough estimate.
Let's pay attention to two circumstances. Firstly, GlacialTech fans are slower, and secondly, they are more efficient. Obviously, this is the result of using an impeller with a more complex blade shape: even at the same speed, the GlacialTech fan carries more air than the Titan: see graph growth. BUT the noise level at the same speed is approximately equal to: the proportion is observed even for fans of different manufacturers with different impeller shapes.
It should be understood that the real noise characteristics of a fan depend on its technical design, the pressure generated, the volume of air pumped, on the type and shape of obstacles in the way of air flows; that is, on the type of computer case. Since there are many different enclosures used, it is not possible to directly apply the measurements measured in ideal conditions quantitative characteristics of fans - they can only be compared with each other for different models of fans.
Price categories of fans
Consider the cost factor. For example, let's take and in the same online store: the results are entered in the tables above (fans with two ball bearings were considered). As you can see, the fans of these two manufacturers belong to two different classes: GlacialTech operate at lower speeds, so they make less noise; at the same speed they are more efficient than Titan - but they are always more expensive by a dollar or two. If you need to build the least noisy cooling system (for example, for a home computer), you will have to fork out for more expensive fans with complex blade shapes. In the absence of such strict requirements or with a limited budget (for example, for an office computer), simpler fans will do just fine. different type The impeller suspension used in fans (for more details, see section ) also affects the cost: the fan is more expensive, the more complex bearings are used.
The connector key is beveled corners on one side. The wires are connected as follows: two central ones - "ground", common contact (black wire); +5 V - red, +12 V - yellow. To power the fan through the molex connector, only two wires are used, usually black ("ground") and red (supply voltage). By connecting them to different pins of the connector, you can get different fan speeds. A standard voltage of 12V will run the fan at normal speed, a voltage of 5-7V provides about half the rotation speed. Preferably use more high voltage, since not every electric motor is able to reliably start at too low a supply voltage.
As experience shows, fan speed when connected to +5 V, +6 V and +7 V is approximately the same(with an accuracy of 10%, which is comparable to the accuracy of measurements: the rotation speed is constantly changing and depends on many factors, such as air temperature, the slightest draft in the room, etc.)
I remind you that the manufacturer guarantees the stable operation of their devices only when using standard voltage nutrition. But, as practice shows, the vast majority of fans start up perfectly even at low voltage.
The contacts are fixed in the plastic part of the connector with a pair of folding metal "antennae". It is not difficult to remove the contact by pressing down the protruding parts with a thin awl or a small screwdriver. After that, the "antennae" must again be unbent to the sides, and insert the contact into the corresponding socket of the plastic part of the connector:
Sometimes coolers and fans are equipped with two connectors: a molex connected in parallel and a three- (or four-) pin. In this case you need to connect power only through one of them:
In some cases, not one molex connector is used, but a pair of "mom-dad": this way you can connect the fan to the same wire from the power supply that powers the hard drive or optical drive. If you are rearranging the pins in the connector to get a non-standard voltage on the fan, please refer to Special attention to rearrange the contacts in the second connector in exactly the same order. Failure to do so will result in the wrong voltage being supplied to the hard drive or optical drive, which will most likely result in their immediate failure.
In three-pin connectors, the installation key is a pair of protruding guides on one side:
The mating part is located on the contact pad; when connected, it enters between the guides, also acting as a retainer. The corresponding connectors for powering the fans are located on the motherboard (usually several pieces in different places on the board) or on the board of a special controller that controls the fans:
In addition to the "ground" (black wire) and +12 V (usually red, less often: yellow), there is also a tachometric contact: it is used to control the fan speed (white, blue, yellow or green wire). If you do not need the ability to control the fan speed, then this contact can be omitted. If the fan is powered separately (for example, via a molex connector), it is permissible to connect only the speed control contact and a common wire using a three-pin connector - this scheme is often used to monitor the fan speed of the power supply, which is powered and controlled by the internal circuits of the PSU.
Four-pin connectors appeared relatively recently on motherboards with processor sockets LGA 775 and socket AM2. They differ in the presence of an additional fourth contact, while being fully mechanically and electrically compatible with three-pin connectors:
Two identical fans with three-pin connectors can be connected in series to one power connector. Thus, each of the electric motors will have 6 V of supply voltage, both fans will rotate at half speed. For such a connection, it is convenient to use fan power connectors: the contacts can be easily removed from the plastic case by pressing the fixing “tab” with a screwdriver. The connection diagram is shown in the figure below. One of the connectors connects to the motherboard as usual: it will provide power to both fans. In the second connector, using a piece of wire, you need to short-circuit two contacts, and then insulate it with tape or electrical tape:
It is strongly not recommended to connect two different electric motors in this way.: due to the inequality of electrical characteristics in various operating modes (startup, acceleration, stable rotation), one of the fans may not start at all (which is fraught with failure of the electric motor) or require an excessively high current to start (it is fraught with failure of the control circuits).
Often, fixed or variable resistors connected in series in the power circuit are used to limit the fan speed. By changing the resistance of the variable resistor, you can adjust the rotation speed: this is how many manual fan speed controllers are arranged. When designing such a circuit, it must be remembered that, firstly, the resistors heat up, dissipating part of the electrical power in the form of heat - this does not contribute to more efficient cooling; secondly, the electrical characteristics of the electric motor in various operating modes (starting, acceleration, stable rotation) are not the same, the resistor parameters must be selected taking into account all these modes. To select the parameters of the resistor, it is enough to know Ohm's law; you need to use resistors designed for a current no less than the electric motor consumes. However, I personally do not welcome manual control of cooling, as I believe that a computer is quite a suitable device to control the cooling system automatically, without user intervention.
Fan monitoring and control
Most modern motherboards allow you to control the speed of fans connected to some three- or four-pin connectors. Moreover, some of the connectors support software control of the speed of rotation of the connected fan. Not all connectors on the board provide such capabilities: for example, the popular Asus A8N-E motherboard has five connectors for powering fans, only three of them support rotation speed control (CPU, CHIP, CHA1), and only one fan speed control (CPU); Asus P5B motherboard has four connectors, all four support rotation speed control, rotation speed control has two channels: CPU, CASE1 / 2 (the speed of two case fans changes synchronously). The number of connectors with the ability to control or control the speed of rotation does not depend on the chipset or southbridge used, but on the specific model of the motherboard: models from different manufacturers may differ in this regard. Often, motherboard designers deliberately deprive cheaper models of fan speed control capabilities. For example, the Asus P4P800 SE motherboard for Intel Pentiun 4 processors is able to regulate the speed of the processor cooler, while its cheaper version Asus P4P800-X is not. In this case, you can use special devices that are able to control the speed of several fans (and usually provide for the connection of a number of temperature sensors) - there are more and more of them on the modern market.
Fan speeds can be controlled using BIOS Setup. As a rule, if the motherboard supports changing the fan speed, here in the BIOS Setup you can configure the parameters of the speed control algorithm. The set of parameters is different for different motherboards; usually the algorithm uses the readings of thermal sensors built into the processor and motherboard. There are a number of programs for various operating systems that allow you to control and adjust the speed of fans, as well as monitor the temperature of various components inside the computer. Manufacturers of some motherboards bundle their products with proprietary programs for Windows: Asus PC Probe, MSI CoreCenter, Abit µGuru, Gigabyte EasyTune, Foxconn SuperStep, etc. Several universal programs are distributed, among them: (shareware, $20-30), (distributed free of charge, not updated since 2004). The most popular program of this class is:
These programs allow you to monitor a number of temperature sensors that are installed in modern processors, motherboards, video cards and hard drives. The program also monitors the rotation speed of fans that are connected to motherboard connectors with appropriate support. Finally, the program is able to automatically adjust the speed of the fans depending on the temperature of the observed objects (if the motherboard manufacturer has implemented hardware support for this feature). In the figure above, the program is configured to control only the processor fan: at a low CPU temperature (36°C), it rotates at a speed of about 1000 rpm, which is 35% of the maximum speed (2800 rpm). Setting up such programs comes down to three steps:
- determining which of the channels of the motherboard controller are connected to fans, and which of them can be controlled by software;
- specifying which temperatures should affect the speed of the various fans;
- setting temperature thresholds for each temperature sensor and operating speed range for fans.
Many programs for testing and fine-tuning computers also have monitoring capabilities:, etc.
Many modern video cards also allow you to adjust the speed of the cooling fan depending on the temperature of the GPU. With the help of special programs, you can even change the settings of the cooling mechanism, reducing the noise level from the video card in the absence of load. This is how the optimal settings for the HIS X800GTO IceQ II video card look in the program:
Passive coolingPassive cooling systems are called those that do not contain fans. Individual computer components can be content with passive cooling, provided that their heatsinks are placed in sufficient airflow created by "foreign" fans: for example, a chipset chip is often cooled by a large heatsink located near the CPU cooler. Passive cooling systems for video cards are also popular, for example:
Obviously, the more heat sinks one fan has to blow through, the more flow resistance it needs to overcome; thus, with an increase in the number of radiators, it is often necessary to increase the speed of rotation of the impeller. It is more efficient to use a lot of low-speed large-diameter fans, and passive cooling systems are preferably avoided. Despite the fact that passive heatsinks for processors, video cards with passive cooling, even power supplies without fans (FSP Zen) are produced, trying to build a computer without fans at all from all these components will certainly lead to constant overheating. Because a modern high-performance computer dissipates too much heat to be cooled only by passive systems. Due to the low thermal conductivity of air, it is difficult to organize effective passive cooling for the entire computer, except to turn the entire computer case into a radiator, as is done in:
Compare the case-radiator in the photo with the case of a conventional computer!Perhaps, completely passive cooling will be enough for low-power specialized computers (for Internet access, for listening to music and watching videos, etc.)
In the old days, when the power consumption of processors had not yet reached critical values - a small radiator was enough to cool them - the question "what will the computer do when nothing needs to be done?" It was solved simply: while it is not necessary to execute user commands or running programs, the OS gives the processor a NOP command (No OPeration, no operation). This command causes the processor to perform a meaningless, ineffectual operation, the result of which is ignored. This takes not only time, but also electricity, which, in turn, is converted into heat. A typical home or office computer, in the absence of resource-intensive tasks, is usually only 10% loaded - anyone can verify this by launching the Windows Task Manager and watching the CPU (Central Processing Unit) Load History. Thus, with the old approach, about 90% of the processor time flew to the wind: the CPU was busy executing commands that no one needed. Newer operating systems (Windows 2000 and later) act more sensibly in a similar situation: using the HLT (Halt, stop) command, the processor is completely stopped at a short time- this obviously allows you to reduce power consumption and processor temperature in the absence of resource-intensive tasks.
Experienced computer scientists can recall whole line programs for "software processor cooling": when running under Windows 95/98/ME, they stopped the processor using HLT, instead of repeating meaningless NOPs, which lowered the temperature of the processor in the absence of computing tasks. Accordingly, the use of such programs under Windows 2000 and newer operating systems is meaningless.
Modern processors consume so much energy (which means: they dissipate it in the form of heat, that is, they heat up) that the developers have created additional technical measures to combat possible overheating, as well as tools that increase the efficiency of saving mechanisms when the computer is idle.
CPU thermal protection
To protect the processor from overheating and failure, the so-called thermal throttling is used (usually not translated: throttling). The essence of this mechanism is simple: if the processor temperature exceeds the allowable one, the processor is forcibly stopped by the HLT command so that the crystal has the opportunity to cool down. In early implementations of this mechanism, through BIOS Setup, it was possible to configure how much time the processor would be idle (CPU Throttling Duty Cycle: xx%); new implementations "slow down" the processor automatically until the temperature of the crystal drops to an acceptable level. Of course, the user is interested in the fact that the processor does not cool down (literally!), but does useful work - for this you need to use a fairly efficient cooling system. You can check if the processor thermal protection mechanism (throttling) is enabled using special utilities, for example:
Minimization of energy consumption
Almost all modern processors support special technologies to reduce energy consumption (and, accordingly, heating). Different manufacturers call such technologies differently, for example: Enhanced Intel SpeedStep Technology (EIST), AMD Cool'n'Quiet (CnQ, C&Q) - but they work, in fact, the same way. When the computer is idle and the processor is not loaded with computing tasks, the clock frequency and voltage of the processor decreases. Both of these reduce the power consumption of the processor, which in turn reduces heat dissipation. As soon as the processor load increases, the full speed of the processor is automatically restored: the operation of such a power saving scheme is completely transparent to the user and running programs. To enable such a system, you need:
- enable the use of supported technology in BIOS Setup;
- install the appropriate drivers in the OS you are using (usually this is a processor driver);
- in the Windows Control Panel, in the Power Management section, on the Power Schemes tab, select the Minimal Power Management scheme from the list.
For example, for an Asus A8N-E motherboard with a processor, you need ( detailed instructions are given in the User's Guide):
- in BIOS Setup, in the Advanced > CPU Configuration > AMD CPU Cool & Quiet Configuration section, switch the Cool N "Quiet parameter to Enabled; and in the Power section, switch the ACPI 2.0 Support parameter to Yes;
- install ;
- see above.
You can check that the processor frequency is changing using any program that displays the processor clock speed: from specialized types, up to the Windows Control Panel (Control Panel), section System (System):
Often, motherboard manufacturers additionally complete their products with visual programs that clearly demonstrate the operation of the mechanism for changing the frequency and voltage of the processor, for example, Asus Cool&Quiet:
The processor frequency changes from maximum (in the presence of computational load) to some minimum (in the absence of CPU load).
RMClock utility
During the development of a set of programs for complex testing of processors, (RightMark CPU Clock / Power Utility) was created: it is designed to monitor, configure and manage the power-saving capabilities of modern processors. The utility supports all modern processors and a variety of power consumption management systems (frequency, voltage ...). The program allows you to monitor the occurrence of throttling, changes in the frequency and voltage of the processor. Using RMClock, you can configure and use everything that standard tools allow: BIOS Setup, power management by the OS using the processor driver. But the possibilities of this utility are much broader: with its help, you can configure a number of parameters that are not available for configuration in a standard way. This is especially important when using overclocked systems, when the processor runs faster than the nominal frequency.
Video card auto overclocking
A similar method is used by video card developers: the full power of the GPU is needed only in 3D mode, and a modern graphics chip can cope with a desktop in 2D mode even at a reduced frequency. Many modern video cards are tuned so that the graphics chip serves the desktop (2D mode) with reduced frequency, power consumption and heat dissipation; accordingly, the cooling fan spins more slowly and makes less noise. The video card only starts to work at full capacity when running 3D applications, such as computer games. Similar logic can be implemented programmatically, using various utilities for fine-tuning and overclocking video cards. For example, this is how the automatic overclocking settings in the program for the HIS X800GTO IceQ II video card look like:
Quiet computer: myth or reality?From the user's point of view, a sufficiently quiet computer will be considered such, the noise of which does not exceed the ambient background noise. During the day, taking into account the noise of the street outside the window, as well as the noise in the office or at work, it is permissible for the computer to make a little more noise. A home computer that is planned to be used around the clock should be quieter at night. As practice has shown, almost any modern powerful computer can be made to work quite quietly. I will describe a few examples from my practice.
Example 1: Intel Pentium 4 platform
My office uses 10 3.0 GHz Intel Pentium 4 computers with standard CPU coolers. All machines are assembled in inexpensive Fortex cases priced up to $30, Chieftec 310-102 power supplies (310 W, 1 80×80×25 mm fan) are installed. In each case, a 80x80x25 mm fan (3000 rpm, noise 33 dBA) was installed on the back wall - they were replaced by fans with the same performance 120x120x25 mm (950 rpm, noise 19 dBA) ). In the file server of the local network, for additional cooling of hard drives, 2 fans 80 × 80 × 25 mm are installed on the front wall, connected in series (speed 1500 rpm, noise 20 dBA). Most computers use the Asus P4P800 SE motherboard, which is able to regulate the speed of the processor cooler. Two computers have cheaper Asus P4P800-X boards, where the cooler speed is not regulated; to reduce noise from these machines, the CPU coolers have been replaced (1900 rpm, 20 dBA noise).
Result: computers are quieter than air conditioners; they are almost inaudible.
Example 2: Intel Core 2 Duo Platform
A home computer based on a new Intel Core 2 Duo E6400 (2.13 GHz) processor with a standard processor cooler was assembled in an inexpensive $25 aigo case, a Chieftec 360-102DF power supply (360 W, 2 80 × 80 × 25 mm fans) was installed. There are 2 fans 80×80×25 mm connected in series in the front and rear walls of the case (speed adjustable, from 750 to 1500 rpm, noise up to 20 dBA). Used motherboard Asus P5B, which is able to regulate the speed of the CPU cooler and case fans. A video card with a passive cooling system is installed.
Result: the computer makes such a noise that during the day it is not audible over the usual noise in the apartment (conversations, steps, the street outside the window, etc.).
Example 3: AMD Athlon 64 Platform
My home computer with an AMD Athlon 64 3000+ (1.8 GHz) processor was built in an inexpensive Delux case priced under $30, initially containing a CoolerMaster RS-380 power supply (380 W, 1 fan 80 × 80 × 25 mm) and a GlacialTech SilentBlade video card GT80252BDL-1 connected to +5 V (about 850 rpm, less than 17 dBA noise). The Asus A8N-E motherboard is used, which is able to regulate the speed of the processor cooler (up to 2800 rpm, noise up to 26 dBA, in idle mode the cooler rotates about 1000 rpm and noise is less than 18 dBA). The problem with this motherboard: cooling of the nVidia nForce 4 chipset chip, Asus installs a small 40x40x10 mm fan with a rotation speed of 5800 rpm, which whistles quite loudly and unpleasantly (in addition, the fan is equipped with a sleeve bearing that has a very short life) . To cool the chipset, a cooler for video cards with a copper radiator was installed; against its background, clicks of the positioning of the hard disk heads are clearly audible. A working computer does not interfere with sleeping in the same room where it is installed.
Recently, the video card was replaced by HIS X800GTO IceQ II, for the installation of which it was necessary to modify the chipset heatsink: bend the fins so that they do not interfere with the installation of a video card with a large cooling fan. Fifteen minutes of work with pliers - and the computer continues to work quietly even with a fairly powerful video card.
Example 4: AMD Athlon 64 X2 Platform
A home computer based on an AMD Athlon 64 X2 3800+ processor (2.0 GHz) with a processor cooler (up to 1900 rpm, noise up to 20 dBA) is assembled in a 3R System R101 case (2 fans 120 × 120 × 25 mm are included, up to 1500 rpm, installed on the front and rear walls of the case, connected to the standard monitoring and automatic fan control system), FSP Blue Storm 350 power supply (350 W, 1 fan 120 × 120 × 25 mm) is installed. A motherboard was used (passive cooling of the chipset microcircuits), which is able to regulate the speed of the processor cooler. Used graphics card GeCube Radeon X800XT, cooling system replaced by Zalman VF900-Cu. A hard drive was chosen for the computer, known for its low noise level.
Result: The computer is so quiet that you can hear the sound of the hard drive motor. A working computer does not interfere with sleeping in the same room where it is installed (the neighbors behind the wall are talking even louder).
This is the company's own development. Fans with 112 mm impeller are equipped with PWM control, thanks to which they can change their speed in the range from 800 to 1800 rpm, creating an airflow of 23.0-68.5 CFM, static pressure of 0.39-2.07 mm H 2 O and noise level 21.9-27.6 dBA.
Under the metal plate on the 41 mm fan stator is a branded UFB (Updraft Floating Balance) bearing with a claimed service life of 150,000 hours, or more than 12 years of continuous operation.
The electrical characteristics of the "turntables" are also on the level: according to our measurements, each fan consumes no more than 1.8 W and starts at 4 V. The length of the four-wire braided cables is 400 mm.
As anti-vibration dampers, silicone rings are inserted into the holes for mounting the fans, and the fastening itself is carried out using wire brackets and plastic studs with holes for these brackets.
The main thing is to correctly install the fans on the radiator, so that one of them works for blowing in, and the second for blowing air out of the radiator.
As for the installation procedure, the fully universal Phanteks PH-TC12DX is fixed to the processor of the LGA2011 construct quite quickly and with just one Phillips screwdriver. But first, threaded support studs are screwed into the mounting holes.
And only then to the guides screwed to these studs, a clamping bar with two spring-loaded screws attracted cooler.
The clamping force is very high, so that the heatsink does not move or rotate on the processor.
In terms of compatibility with high heat sinks on memory or power elements, the situation is twofold. It would seem that the distance from the board to the lower edge of the fans is 48 mm, which is not enough for memory modules with comb heatsinks that have been fashionable lately.
However, let's remind you that the cooler is relatively narrow, so if it does block memory slots, then only one or two closest to the processor socket - and nothing more.
The height of Phanteks PH-TC12DX will fit even in relatively narrow cases, since after installation on the processor it turns out to be no higher than 165 mm.
Let's see what new will please us today's competitor Phanteks PH-TC12DX.
⇡ Thermaltake NiC C5 (CLP0608)
As we already mentioned in the introduction of today's article, Thermaltake released four coolers of the new NiC line at once. Model C5 (CLP0608) is the oldest and most expensive of them. A series of coolers of the NiC series (Non-interference Cooler - in the literal translation "non-interference cooler") is designed specifically for systems with memory modules equipped with high heatsinks, which have recently become very popular.
The box, made of thick cardboard, is no less informative than Phanteks. Here are technical specifications, and a description of key features with photos, and a list of supported platforms.
|
|
|
|
Inside the cardboard box there are soft polyurethane inserts in the form of a cooler in which it is fixed. Accessories are sealed in a separate box. These include steel rails and a set of fasteners, a plastic reinforcing plate, as well as instructions and thermal paste.
The Thermaltake NiC C5 costs $5 more than Phanteks, which is $55. The cooling system comes with a three-year warranty. The country of production is China.
Thermaltake NiC C5 is a medium-sized bright and eye-catching cooler. The red fan frames contrast with the black impellers and black plastic "shells" that cover the heatsink.
It is simply impossible not to pay attention to such a cooler. Its height is 160mm, width is 148mm, and its thickness is only 93mm, which is really not much for a cooler with two fans.
The fans are mounted on blow-out and fixed in plastic shells that leave the sides of the radiator open ...
|
|
|
|
…as well as its top and bottom in the heatpipe areas.
The radiator itself is assembled with 52 aluminum plates 0.4 mm thick, pressed onto heat pipes with an intercostal distance of 1.7 mm.
The area of such a radiator is slightly larger than that of Phanteks PH-TC12DX - it is 5780 cm 2 .
Five six-millimeter nickel-plated heat pipes are soldered to the base in grooves, in which they are laid without gaps.
Nickel-plated copper plate with dimensions of 40x40 mm and minimum thickness 1.5 mm (under tubes) perfectly polished.
However, unlike the Phanteks blade, its evenness leaves much to be desired. The bulge in the center of the base did not fail to affect the usefulness of the contact between the cooler heatsink and the processor heat spreader.
Two 120x120x25 mm fans rotate synchronously and are equipped with a speed controller.
It is installed on a short cable extending from the three-pin connector for connecting fans to the motherboard.
In our opinion, this method of adjustment is inconvenient, since to change the fan speed each time you have to open the case of the system unit. As for the fans themselves, they are interesting in the shape of the blades, consisting of two sail-shaped halves.
In the description of Thermaltake NiC C5, this solution is not explained in any way, which is strange, because marketers love such “features” so much. In our opinion, these blades are made to increase the pressure of the air flow pumped between the radiator fins, because NiC C5 turned out to be relatively dense.
The fan speed can be adjusted from 1000 to 2000 rpm. Maximum airflow is claimed at 99.1 CFM, static pressure is 2.99 mm H 2 O, and noise levels should range from 20 to 39.9 dBA.
The sticker on the 40mm stator shows the name of the fan model and its electrical specifications.
With 3.8 W for each turntable declared in the specifications, one fan consumed a little more than 4 W, which is twice as much as Phanteks. But the starting voltage turned out to be slightly lower - 3.8 V. Cable length - 300 mm. The bearing is conventional - sliding, with a standard service life of 40,000 hours, or more than 4.6 years of continuous operation.
The procedure for installing NiC C5 is detailed in the instructions, but in our case - for a platform with an LGA2011 connector - it is no different from installing Phanteks PH-TC12DX.
After installation on the board, the distance to the lower border of Thermaltake NiC C5 is only 36 mm.
However, as we mentioned above, it at the same size as most other dual fan coolers, so it's unlikely to get in the way of installing RAM modules with tall heatsinks.
In terms of height, Thermaltake is only 3 mm higher than Phanteks, therefore, most likely, it will also fit in narrow cases of system units without any problems.
Well, it looks, in our opinion, more attractive. However, the taste and color, as they say ...
⇡ Test configuration, tools and testing methodology
Testing of cooling systems was carried out in a closed case of the system unit of the following configuration:
- Motherboard: Intel Siler DX79SR (Intel X79 Express, LGA2011, BIOS 0559 from 03/05/2013);
- CPU: Intel Core i7-3970X Extreme Edition 3.5-4.0 GHz(Sandy Bridge-E, C2, 1.1V, 6x256KB L2, 15MB L3);
- Thermal interface: ARCTIC MX-4 ;
- RAM: DDR3 4x8GB G.SKILL TridentX F3-2133C9Q-32GTX (2133MHz, 9-11-11-31, 1.6V);
- Video card: AMD Radeon HD 7770 GHz Edition 1GB GDDR5 128bit 1000/4500MHz (with Deepcool V4000 passive copper heatsink);
- System Drive: 256 GB Crucial m4 SSD (SATA-III, CT256M4SSD2, BIOS v0009);
- Drive for programs and games: Western Digital VelociRaptor (SATA-II, 300 GB, 10000 rpm, 16 MB, NCQ) in a Scythe Quiet Drive 3.5″ box;
- Backup disk: Samsung Ecogreen F4 HD204UI (SATA-II, 2 TB, 5400 rpm, 32 MB, NCQ);
- Frame: Antec Twelve Hundred(front wall - three Noiseblocker NB-Multiframe S-Series MF12-S2 at 1020 rpm; back - two Noiseblocker NB-BlackSilentPRO PL-1 at 1020 rpm; top - standard 200 mm fan at 400 rpm) ;
- Control and monitoring panel: Zalman ZM-MFC3 ;
- Power supply: Corsair AX1200i (1200W), 120mm fan.
For basic tests, a six-core processor at a reference frequency of 100 MHz with a fixed multiplier of 44 and activated Load-Line Calibration was overclocked to 4,4 GHz with increasing voltage in the BIOS of the motherboard to 1.245~1.250V. Turbo Boost technology was turned off during testing, but Hyper-Threading was activated to increase heat dissipation. The voltage of the RAM modules was fixed at around 1.6 V, and its frequency was 2.133 GHz with timings of 9-11-11-31. Other BIOS settings related to overclocking the processor or RAM were not changed.
Testing carried out in operating system Microsoft Windows 7 Ultimate x64 SP1. The software used for the test is as follows:
- LinX AVX Edition v0.6.4 - to create a load on the processor (allocated memory - 4500 MB, Problem Size - 24234, two cycles of 11 minutes each);
- Real Temp GT v3.70 - for monitoring the temperature of the processor cores;
- Intel Extreme Tuning Utility v4.0.6.102 - for monitoring and visual control of all system parameters during overclocking.
A full screenshot during one of the testing cycles looks like this:
The load on the processor was created by two consecutive LinX AVX cycles with the above settings. It took 8-10 minutes to stabilize the processor temperature between cycles. The final result, which you will see in the diagram, is the maximum temperature of the hottest of the six CPU cores at peak load and in idle mode. In addition, temperatures of all processor cores and their average values will be shown in a separate table. The room temperature was controlled by an electronic thermometer installed next to the system unit with a measurement accuracy of 0.1 ° C and the possibility of hourly monitoring of changes in the temperature in the room over the past 6 hours. During this test, the ambient temperature was unusually high, as the summer heat set in outside the window - it fluctuated in the range 27,6-28,0 °C
The noise level of the cooling systems was measured using an electronic sound level meter CENTER-321 in the period from one to three in the morning in a completely closed room of about 20 m 2 with double-glazed windows. The noise level was measured outside the case of the system unit, when the source of noise in the room was only the cooler itself and its fan. The sound level meter, fixed on a tripod, was always located strictly at one point at a distance of exactly 150 mm from the fan stator. The cooling systems were placed at the very corner of the table on a polyurethane foam substrate. The lower limit of sound level meter measurements is 29.8 dBA, and the subjectively comfortable (please don't confuse it with low!) cooler noise level when measured from such a distance is about 36 dBA. The fan speed was varied throughout the entire range of their operation using a special controller by changing the supply voltage in 0.5 V steps. Test results and their analysis
Cooling efficiency
The results of testing the efficiency of cooling systems are presented in the table and in the diagram:
To put it bluntly, both novelties did not impress us with their effectiveness. Thermaltake NiC C5 is able to demonstrate the same efficiency as the legendary Thermalright TRUE Spirit 140, but only at high speeds of its two fans and, of course, yielding to TRUE Spirit 140 in noise level. At a quiet 800 rpm, the efficiency of the NiC C5 is rather mediocre - in this mode, it loses TRUE Spirit 140 immediately by 4 degrees Celsius in terms of peak processor temperature. As for the Phanteks PH-TC12DX, unlike its older brother, this is an even less efficient cooling system. For example, at the maximum speed of its two fans, Phanteks demonstrates the same efficiency as the cheaper TRUE Spirit 140 with one fan at 800 rpm. And at 800 rpm, the PH-TC12DX did not cope with the cooling of the overclocked processor at all, as, indeed, at 1000 rpm. We understand that the ambient temperature during these tests was relatively high, however, in the summary chart, where all results are given at an ambient temperature of 25 degrees Celsius, Phanteks PH-TC12DX and Thermaltake NiC C5 do not shine with efficiency. It is to this that we now turn.
Let's add the results to the summary table* and to the diagram, where all the tested coolers are presented in their standard configurations in quiet mode and at maximum fan(s) speed when the processor is overclocked to 4.4 GHz and voltage is 1.245~1.250 V:
* The peak temperature of the hottest processor core is shown in the diagram taking into account the delta from room temperature and for all cooling systems is reduced to 25 degrees Celsius.
Thermaltake NiC C5 at the maximum speed of two fans managed to take its place in the middle group of coolers, but its noise level is the highest in it. In quiet mode at 800 rpm, this model is only the fourth from the end. In turn, the even less efficient Phanteks PH-TC12DX is the leader in the third group of coolers, though only in terms of noise level, while losing out in efficiency to Noctua NH-U14S and the same Thermalright TRUE Spirit 140 at 800 rpm. Yes, and with a huge difference in noise level.
It is logical that with such efficiency it is pointless to talk about further overclocking of the processor when it is cooled by Phanteks PH-TC12DX, but Thermaltake NiC C5 allowed the Intel Core i7-3970X Extreme Edition to maintain stability at a frequency of 4600 MHz at a voltage of 1.3 V and a peak temperature of the most hot core 84 degrees Celsius:
Thus, if you do not pay attention to the high noise level, Thermaltake NiC C5 looks quite confident in our "Table of Ranks" with the maximum overclocking of the processor.
Well, Phanteks PH-TC12DX leads the top three coolers with basic overclocking of the processor, yielding to two brothers in misfortune - Deepcool Ice Blade Pro and Noctua NH-U12S - in terms of noise level. We now turn to the evaluation and analysis of the latter.
Noise level
The noise level of the participants in our today's tests was measured over the entire range of operation of their fans according to the method described in the corresponding section of the article and is presented in the graph:
In short, both novelties are noisy. It's not so much a significant loss compared to the Thermalright TRUE Spirit 140 with a single fan, but the noisy pairs of Phanteks PH-TC12DX and Thermaltake NiC C5 fans themselves. This is especially true for the Thermaltake model, which stands out not only for the characteristic resonance of the operation of fans installed for intake and exhaust, but also for the uneven change in their noise depending on the speed, which is clearly seen from the broken curve. The Phanteks PH-TC12DX is superior in this regard, remaining comfortable at around 950 rpm, while the Thermaltake NiC C5 is comfortable at 890 rpm. Both novelties can be called quiet only if the speed of their fans does not exceed 800 rpm.
⇡ Conclusion
Both of the new dual-fan coolers that we reviewed and tested today failed to please us with either outstanding efficiency or low noise levels. The Thermaltake NiC C5 from this pair is more efficient, but it looks rather pale in comparison with the mass of other air coolers, including more affordable ones. The Phanteks PH-TC12DX is quieter, but it's really quiet only at speeds when it can no longer handle even moderate overclocking of a six-core processor. The Thermaltake NiC C5 fans are equipped with a manual stepless controller on a short and uncomfortable cable, while the Phanteks PH-TC12DX has PWM control. Also of the differences, we note the mirror base of Thermaltake, a small difference in cost, more durable and economical fans, as well as a 7 mm higher fit over the board in favor of Phanteks. Otherwise, these coolers are the same. They are versatile, easy to install, and each of them looks attractive in its own way. But whether these pluses are enough and whether you choose one of them to cool the processor is up to you.
