New coolers Noctua NH-U9 and NH-U12 - quiet and effective cooling
Author: Date: 23.05.2006 |
|
Benchmarking
Test bench configuration and benchmarking specifics
It is quite evident that this time we received representatives of the Hi-End cooling systems, and we'll be running extensive tests to the full. We selected the most serious rivals:
- Zalman CNPS9500LED - the most powerful and product by the eminent Korean manufacturer, is one of the most popular coolers among enthusiasts and fanciers of produce manufactured under this brand name. Its design has consolidates the latest engineering advancements, so Zalman CNPS9500LED is indeed a serious opponent to heroes of today's review.
- Scythe Ninja - the second most popular cooler for enthusiasts, stands out with its impressive size and exceptional performance. Besides, the cooler can be used in the passive mode - without a fan.
- GlacialTech Igloo 5700MC/7700MC - new super coolers by GlacialTech have already demonstrated their might, and in view of the affordable price both the coolers are confident at conquering a place of dignity in the computer of the Russian PC enthusiast.
All specialists know that on Intel and AMD platforms coolers behave differently, so to run full-featured and impartial tests we assembled two different platforms:
Test bench configuration for the AMD platform
|
Processor
|
AMD Athlon 64 3200+ (Venice, 2.0 GHz, 512K L2 cache, revision E6)
|
Motherboard
|
DFI LanParty NF4-D (nForce 4)
|
RAM
|
2 x 512 DDR 500 Kingmax HardCore PC4000
|
Video Card
|
256 MB Sparkle NVIDIA GeForce 7600GT
|
HDD
|
120 GB Seagate SATA II, 8 MB cache (ST3120813AS)
|
Housing
|
ThermalTake Xaser III (window, 4 housing fans - 80 mm)
|
PSU
|
FSP Optima 600W (OPS600-80GLN)
|
AMD Athlon 64 3200+ is probably not the hottest processor in the AMD line, but it is just this processor that has gained maximum occurrence because of its low price and superb overclocking capability.
Despite the reduced RPM of the fans, the housing of ThermalTake Xaser III provides excellent ventilation inside the housing and hot air is not kept inside, so the conditions are close to ideal.
We ran tests on the Intel platform in the same housing.
Test bench configuration for the Intel platform
|
Processor
|
Intel Pentium D 930 (Presler, Dual Core 3.0 GHz, 4 MB L2 cache, revision B1)
|
Motherboard
|
ASUS P5LD2 (i945P)
|
RAM
|
2 x 512 DDR2 Samsung PC5300
|
Video Card
|
256 MB Sparkle NVIDIA GeForce 7600GT
|
HDD
|
120 GB Seagate SATA II, 8 MB cache (ST3120813AS)
|
Housing
|
ThermalTake Xaser III (window, 4 housing fans - 80 mm)
|
PSU
|
FSP Optima 600W (OPS600-80GLN)
|
To verify the capabilities, we managed to get the latest dual-core Intel Pentium D 930 (3.0 GHz) based on the 65-nm Presler core of revision B1. This processor is among Intel's "hottest" processors and offers the nominal heat package 95 W, but after overclocking the coolers would have to dissipate much more heat, and even 150 W might seem a magic.
Now regarding the overclocking. It is absolutely funny to test Hi-End coolers with AMD Athlon 64 3200+ in the nominal mode because any cooler will cope with its heat emission of no more than 67 W. So we decided not to waste time in vain, and verified coolers on the AMD platform with the processor overclocked to 2.7 GHz, whose voltage was raised from 1.4 to 1.55 V. That is pretty standard overclocking for AMD Athlon 64 processors (Venice core).
We had to mess around with the Intel platform for quite a while - we received not the most outstanding processor specimen, so we were unable to overclock it to clock speeds higher than 3.9 GHz. But in any case it is already an impressive overclocking because batch-produced Intel Pentium D processors run at no more than 3.4 GHz (model Intel Pentium D 950). To attain 3.9 GHz, we had to raise the voltage from 1.275 to 1.400 V.
The benchmarking methodology is simple enough. The performance of coolers was verified in the three most popular modes:
- Simple. In this mode, the computer was simply left idle for about 15-20 minutes until the temperature stabilized. In this mode the computer is used for text-editing tasks, watching films, surfing over the Internet, listening to the music, etc.
- 3D games. This is rightfully the second most popular mode of computer utilization. No doubt, different games load the processor in different ways, but that is not critical. To verify the coolers in this mode, we launched 20x run of the "Troll`s Lair", a test from the well-known 3D Mark 2003 benchmarking suite.
- Max load. This is more likely a "strength test", and to generate maximum load on the processors we used the latest version of the known utility S&M (v.1.8.0 alpha). There has been repeatedly mentioned that it is able heating up the processor much stronger than any working applications, so we call this mode the "strength test".
It's no good using the cooler at maximum RPM of the fan, especially when there is SUCH a cooler on the processor. So all the participants of tests were verified in the two modes: at the maximum RPMs of the fan and at a speed reduced to the noiseless level.
Tests were run in the ordinary room environment where the air temperature was about 23 C. For convenience, we took the ambient temperature as the reference point on the graphs.
Well, let's get round to business!
AMD platform
Frankly, at the beginning it was even hard to believe that during transition from the idle mode to gaming tests the temperature wouldn't go up at all(!), but that was a fact. Only in some cases the temperature was raised by 1 degree. It is rather hard to talk about leaders and outsiders here because none of the coolers allowed the processor to heat up to even 50 C, but it was easy to see that Noctua NH-U12 showed the best result and left all the competitors at least by 1 degree. So, we register one victory in favor of Noctua NH-U12.
Intel platform
At the first stage, we test coolers in the nominal mode of processor operation:
Dual-core Intel processors already in the nominal mode offer a high heat emission which exceeds even overclocked AMD Athlon 64, so the power-saving technologies remained enabled.
We intentionally mentioned above that coolers behave differently on different platforms, and in our case the situation was just like that. The former leader, Noctua NH-U12, gave in its first place to GlacialTech Igloo 5700MC, the most inexpensive and noisy of the four coolers.
The second stage of testing on the Intel platform was decisive. How will the coolers cope with the immense heat emission from the overclocked dual-core CPU by Intel? That's the way they will:
The temperatures went up sharply by 10-15 degrees! Such figures look impressive against the temperature graphs for the AMD platform, although we should make a reservation for the single-core AMD Athlon 64 3200+ CPU.
The overclocking didn't change but fixed the established situation: the leader is GlacialTech Igloo 5700MC as before, and Noctua coolers take the last lines in the rating list.
Nevertheless, under 1-2 C temperature difference the rating looks more formal because all the coolers are good. Only Noctua NH-U9 lags well behind the others.
Interesting is the fact that the temperature difference between the maximum RPMs of the fan and the reduced RPMs for Noctua NH-U12 amounts to merely 1 C albeit at the maximum load. This tells to the low air resistance (good blowing capacity) of the radiator, which is a definite plus. Such a cooler does not need a fan, so it doesn't make any noise.
A bit confusing are the too high temperatures - actually, we expected lower values. The cause of such figures is evident. Have you often noticed the trace of thermo paste left on the cooler base after dismantling the cooling system? If you have seen such a picture, then know that you were unlucky with the processor specimen:
Non-uniformity of the heat-spreading processor lid is a common problem of modern processors.
What catches the eye is that during tests on the AMD platform the Noctua NH-U12 cooler confidently took the first place, and on transition to the Intel platform it dropped to the last but one place. Most likely, that was caused by a combination of the two facts: poor contact with the Intel processor was aggravated by not the best polishing of the base in Noctua coolers. Under the intensive heat emission generated by the "hot" dual-core Intel CPU, the quality of contact between the CPU and the base is a very important parameter. Probably, it is just that which caused such poor results. On the other hand, the 1-3C temperature difference is not critical.
The package bundle of Noctua coolers contains two pair of wire braces to fasten a fan, i.e. it is possible to install not one but two fans on the cooler - one for air suction, the other for air expulsion. Of course, we tried this mode either. For that, we took two quiet 120 mm GlacialTech fans whose rotational speed does not exceed 950 RPM. In view of the extremely low air resistance of Noctua NH-U12, use of two quiet fans may prove quite useful. But practice proved different: the temperature difference reached 1 C only in the "Max load" mode, so the conclusion is self-evident. Use of two fans does not bring a noticeable effect from using a radiator of good blowing capacity, which was just the case we had. A remarkable result!
So, our tests are coming to an end. We have found out the leaders and explored their specifics. We suggest that you should recall this line from the description for Scythe Ninja cooler: "Besides, the cooler can be used in the passive mode - without a fan". And then let's recall that Noctua NH-U12 also offers parameters sufficient for effective use in the passive mode. Shall we try it?
Tests of two giants in the passive mode. Scythe Ninja versus Noctua NH-U12
A cooler for passive cooling should first offer a large area and a distance between the fins sufficient to create a minimum resistance to air flow in the housing. Both rivals offer these qualities to the full. On the other hand, Scythe Ninja has a clear superiority: it offers more heat pipes (not 4 but 6) and the distance between fins is greater (not 3 mm but 5 mm).
Tests were run on the same test benches and under the same conditions. To make it possible for you to imagine all this structure in action, look at the photo of our test bench:
The Scythe Ninja looks simply huge, it takes up almost the whole depth of the housing. But that is just the plus for a passive cooler - air flows in the housing are directed from the front panel where "suction" fans are positioned towards the rear wall with "expulsion" fans. The PSU also pushes air upwards. In the end, it turns out that the major air flow is passed mostly through the radiator of the passive cooler. Noctua NH-U12 is not so wide, but is also on the main channel of air movement because the conditions are practically equal.
Having set standard frequencies on both platforms, we succeeded in attaining the following results:
Both coolers run on par, with Scythe Ninja running at a negligible superiority.
It is absolutely evident that for both monsters the CPU AMD Athlon 64 3200+ is not a load at at all, such temperatures are no more than "fun". But Intel Pentium D has once again proved its "hot" character.
The AMD processor has confidently coped with passive cooling while overclocking it to 2600 MHz (1.55 V), which is a bit lower than before, but anyway excellent. The Intel processor is hot at even the nominal frequency, but we anyway ran the risk to try overclocking - this time without raising voltage. The result proved really impressive - 3600 MHz!
The raise of frequency by 600 MHz has added merely 4-5 C to the temperature of the Intel processor and preserved stability. At the same time, let's not forget that we were able to heat up the CPU to such a temperature only with S&M, because real applications are not able doing such "brutality". So we shouldn't feel any doubts in the stability.
The temperature of AMD Athlon 64 rose quite noticeably, but it is no wonder about that because we already raised voltage, so it is expectable.
It is quite evident that Scythe Ninja is a bit more efficient in the passive mode, which was expectable, its structure was originally calculated for operation in the passive mode, whereas Noctua NH-U12 is "cut out for" operation with a quiet 120 mm fan. Nevertheless, Noctua NH-U12 simply did a fantastic job of fan-free cooling of overclocked processors, and the 3 C lag behind the best passive cooler does not count.
Conclusion
Well, we have come to the end of a review of debut coolers by Noctua. As we can see, engineers at the Austrian Institute of Heat Transmission and Fan Technology have used their skills and abilities to the best. The radiators offer a well thought out design and very low resistance to air flow. Therefore, quiet fans can be used in combination with them without noticeable loss of performance. That is a very important point - Noctua coolers are able cooling efficiently and quietly. There were of course some shortcomings, but so far we have not yet come across an ideal cooler without a single issue. Let's briefly list the major advantages and shortcomings of Noctua's novelties:
Pros:
- Excellent efficiency;
- Compatibility to all the popular platforms;
- Possibility to run effectively with quiet fans;
- High efficiency in the passive mode (for Noctua NH-U12 cooler).
Cons:
- Not completely polished base;
- Lack of fan in the package bundle;
- Dependence of the cooler direction on positioning the processor socket on motherboards for Sockets 754/939/940.
It's a pity that such coolers are not yet available in Russia, but Noctua is actively looking for partners and distributors, so it is quite possible that in a few months we will see the familiar box of cherry color. We appreciate the computer megamarket chain Unicom for the presented and tested hardware.
|
Content: |
|
|
|
Top Stories: |
|
|
|
MoBo:
|
|
|
|
VGA Card:
|
|
|
|
CPU & Memory:
|
|
|