by Sergey Lepilov
04/01/2008 | 10:13 AM
I have recently found out that alternative cooling solutions for graphics cards are not as popular as I expected them to be, according to the head of one large computer service center. It really struck me as very strange because most reference graphics card coolers are not efficient enough and if they are, they often generate too much noise. So why don’t users go for something more advanced that would guarantee better cooling at a more comfortable level of operational noise?
In fact, the answer to this question is pretty simple: most of them are afraid they will lose their graphics card warranty, because not every user can remove the reference cooling system without leaving any marks or ensuring integrity of all stickers. However, in case something happens with the hardware, removed VGA cooler seems like a very good reason to void the warranty and refuse the exchange or return. So, gamers have to put up with high noise and an additional heat source inside of their system case, and when the graphics card warranty expires, the card is no longer powerful enough to satisfy their gaming needs, so purchasing a new cooling system makes simply no sense any more.
However, there is a group of users called overclocking fans who tend to know at least a little more than most mainstream computer users and what is even more important, who know what they need it for. These are the users for whom we write our articles devoted to new graphics card cooling solutions in particular.
Today we would like to introduce to you two newcomers from XIGMATEK and AURAS with highly original and even a little bit funny names. So let’s get started!
Well, heatpipe direct touch technology that we are very well familiar with already from our CPU cooler reviews has finally made it to the graphics card cooling solutions. The first one to bring it on in the VGA market is a cooler from a company we all know very well already – XIGMATEK.
The cooler comes in a small cardboard box with a few cut out windows in the front and the back:
The cooler has a pretty vivid name: BATTLE-AXE VD964, a real battle axe with two blades. As we found out later, there is a very good reason for this symbolic name.
The cooler accessories bundle is packed into a smaller box sitting at the top of the main package. It contains the following items:
Now let’s take a closer look at the cooler.
The size of this newcomer is not as impressive as that of the recently reviewed Arctic Cooling Accelero Xtreme 8800, but it is still pretty big for a graphics card air cooler: it measures 194 x 110 x 32.4mm and weighs 617g. The cooler is really not that small:
The cooler sits on four copper heatpipes 6mm in diameter that form part of the cooler base. The heatpipes carry a heatsink array of numerous aluminum plates. The front of the heatsink is covered with a casing with two 92x25mm fans:
However, I cannot really explain why the heatsink plates are sort of closed at the bottom of the array creating a concave wall for the airflow generated by the fans:
It strikes as a pretty questionable solution for maximum efficiency, because otherwise the airflow from the fans could have cooled down the card PCB and its components.
It turned out that the cooler is actually much taller than the technical specifications claim: it measures 60mm instead of the claimed 32.4mm:
Maybe the manufacturer reported only the height of the heatsink array without taking into account the fans with the casing:
The casing can be easily unscrewed revealing two 92mm fans:
The fans are fastened with four rubber spindles that are inserted between the heatsink plates:
Note that in the center of the heatsink these spindles are not installed at all neither into the heatsink nor into the retention holes on the fans: they are shipped in a separate plastic bag. Looks like they couldn’t figure out a way to insert them there, as it doesn’t seem like an easy task. Besides the rubber jacks, the fans are also held by the casing on top of them, so you shouldn’t worry that they will not be sitting reliably enough.
The fans use “Hypro” bearings and their blades rotate at a constant speed of ~2200RPM generating maximum 28dBA of noise at 40.2CFM airflow:
From my subjective standpoint, the cooler is a little bit noisy. It is a pity that its fans do not support PWM mode, however, you can manually power them from 7V or even 5V source. We are going to check it out later in this review.
The fans can be removed revealing more details of the cooler aluminum heatsink:
There is no base polishing to talk about here, but we definitely have to point out extremely big distance between the heatpipes in the cooler base. Take a look:
The aluminum parts between the cooler heatpipes is a little less than 3mm, so you may get the impression that this pretty big distance may gravely affect the cooler efficiency. You may get even more concerned when you check out the thermal compound imprint on the cooler base…
… and then on the graphics processor heat-spreader:
And if you install the cooler on top of a bare graphics chip, only two heatpipes will actually be working:
I would like to refrain from making any conclusions before the test results come in, but frankly speaking, the heatpipes placed so far apart concern me a lot.
The complete list of supported graphics cards is given in the Technical Specifications section of this review. The cooler can be installed on any of them very quickly and easily. First you fasten two metal plates bundled with the cooler to its base, then you insert the retention spindles into their corresponding holes and then set the graphics card on top of them. After that all you need to do is fasten the graphics card with spring screws at the back of the PCB:
The double battle-axe installed onto a GeForce 8800 GTX looks very serious and impressive:
The massive cooler design has an even greater effect if you look at the graphics card from the top:
I am sure that not everyone will dare install such a massive and heavy (617g) cooling system onto his graphics card. But if you decided to do it, then you should give up the next three PCI slots on your mainboard for good:
It is totally up to you to decide if it is worth it or not, we hope our article will help you to make this decision. In conclusion I would like to say that the recommended retail price for XIGMATEK BATTLE-AXE VD964 is seta round $45.
We are already familiar with AURAS products, but so far we have discussed and tested only CPU coolers from this company. Today we are proud to offer you a detailed review of their new graphics card cooler AURAS Fridge (JES-988).
The cooler arrived in a small flat box with very simple and light-colored design. The front of it bears a large photo of the cooling system inside and has a small transparent window revealing part of the cooler heatsink with heatpipes:
The reverse side of the box lists all supported graphics accelerators and shows the detailed dimensions of the cooling system.
AURAS Fridge comes bundled with the following accessories:
It turned out that AURAS Fridge (stands for “refrigerator”) is a passive heatsink that comes without any fans. Although it doesn’t at all mean that you cannot install any on top of it. The heatsink measures 227.2 x 134.4 x 34mm and weighs approximately 348g. It consists of an aluminum plates array of variable height that sits on four copper heatpipes 6mm in diameter:
The main distinguishing feature of the newcomer is the heatpipes positioning horizontal to the graphics card PCB. In other words, the heatpipes originating from the cooler base go not up or down, but parallel to the graphics card PCB, which allows reducing the cooler height and increase the effective cooling area of its heatsink.
There are total 54 plates in the heatsink array that are spaced out at 4-5mm, which indicates that the new cooling system is primarily designed to work in passive mode:
Without the fans AURAS Fridge turns out quite compact:
Although if you top it with a 25mm fan (or fans) it will be pretty much as tall as the XIGMATEK BATTLE-AXE VD964 we have just discussed.
The heatpipes originating from the cooler base are bent to get as close to the heatsink array as possible and move away from the graphics card PCB components:
The cooler base is made of nickel-plated copper and the heatpipes are slightly flattened to ensure better and bigger contact surface with the base:
The evenness and polish quality of the base plate are impeccable, as you can see.
The cooler can be installed on a pretty wide variety of graphics cards very easily. All you need to do is determine what retention holes to you for your particular graphics card and attach the cooler with four or two screws from the bottom of the card PCB:
When the cooler is installed onto a GeForce 8800 GTX based graphics card it may look pretty bulky, even when used without any fans in the passive mode:
However, the manufacturer designed the cooler capable of accommodating one 120x25mm fan or two 80x25mm fans. To fasten them you should use the bundled plastic clips that lock in the corresponding heatsink retention holes:
This very simple, convenient and reliable solution ensures quick installation of an active fan:
Don’t even try fitting two 120mm fans onto AURAS Fridge heatsink, as the retention holes will not match. Besides, AURAS Fridge is smaller than 240mm (two 12mm fans combined). However, you can easily fit a pair of 80mm fans. In this case you will have to turn one of the retention plates with the retention facing downwards. Moreover, although the manual doesn’t say that the cooler is compatible with a pair of 92mm fans, I could easily fit them both onto the heatsink fastening only at the top (I used the fans from the XIGMATEK cooler).
If you install AURAS Fridge in the passive mode, it will block only one PCI slot next to the graphics card. However with the fans attached, you will lose all three slots next to it:
The recommended retail price for this cooler is $44.99.
The table below sums up all the technical specifications of the new coolers compared with the recently tested Arctic Cooling Accelero Xtreme 8800:
We tested the cooling efficiency of new XIGMATEK and AURAS graphics card coolers in a closed testbed only. The testbed was configured as follows
We put together the following testbed for our experiments:
Our quad-core processor was overclocked from the nominal frequency to 4GHz with the Vcore increased to 1.5825V. DDRII memory worked at 1066MHz frequency and 2.05V voltage.
The testing programs were installed under Windows XP Professional Edition SP2. We used DirectX 9.0c libraries, and ForceWare 171.16 graphics card drivers. The graphics cards were loaded up by running 3DMark06’s Firefly Forest synthetic test ten times in 1920x1200 with 16x anisotropic filtering and 4x FSAA activated. We overclocked the graphics cards and then monitored their temperatures using RivaTuner v2.0.6 utility. We performed at least two test cycles for each cooler and waited for 10-15 minutes for the temperature to stabilize between each test cycle. The ambient temperature remained at ~25°C during the tests and is used as a starting point on our results charts.
The noise level of each cooler was measured according to our traditional method using CENTER-321 electronic noise meter. The subjectively comfortable level of ~34.5dBA is marked with a blue dotted line in the diagrams; the ambient noise from the system case including a processor cooler didn’t exceed ~33.2dBA measured at a 1m distance.
Both graphics cards that will be used for our today’s tests are equipped with a reference cooler by default:
The default coolers didn’t limit the overclocking potential of the graphics cards, however, we couldn’t raise the maximum frequencies any further even with the most efficient cooling solutions of our today’s testing participants. The graphics cards were tested at the following frequencies:
I would like to add that we used the same Arctic Silver 5 thermal compound for all our testing participants.
Besides the reference coolers we have also included the results of one of the today’s best air coolers for GeForce 8800 GTX graphics accelerator – Arctic Cooling Accelero Xtreme 8800, and for GeForce 8800 GTS 512MB – Arctic Cooling Accelero S1 equipped with Turbo-Module. It is not an Arctic Cooling promotion. The thing is that there are hardly any really efficient and at the same time quiet coolers for graphics cards in the market these days.
The cooling efficiency diagram shows the coolers lined up in a descending order for the graphics processor temperature (GPU Temp) and ambient temperature (Ambient Temp) under maximum workload. First let’s check out the results for GeForce 8800 GTS 512MB:
Well, finally there came the time when even a cooler like Arctic Cooling Accelero S1 becomes the least efficient solution (besides the reference cooling systems, of course). Both newcomers from XIGMATEK and AURAS demonstrated excellent cooling efficiency on the overclocked graphics core as well as in terms of the ambient graphics card temperature. Here I have to stress that in case of BATTLE-AXE cooler installed directly onto G92 chip only two heatpipes out of four actually work, because only to have direct connect with the chip. I doubt that aluminum plate gets the other two heatpipes involved into the cooling process, however, despite this fact the XIGMATEK cooler becomes one of the leaders of our today’s test session. I said “one of the leaders”, because the laurels in the most comfortable quiet mode belong indisputably to AURAS solution.
I think expecting AURAS Fridge to cool the overclocked GeForce 8800 GTS 512MB in passive mode would be too much to ask, because it is one of the top chips of Nvidia based graphics accelerators these days, which is also overclocked quite significantly. However, it easily copes with an overclocked GeForce 8600 GTS keeping its temperature at around 70ºC without any fans involved. As for the active mode, I have to say that a pair of two 92-mm fans is definitely more preferable than a single 120-mm fan in terms of maximum cooling efficiency. We checked out two possible positions for the single 120-mm fan: closer to the graphics card output panel and on the other end of the PCB right above the voltage regulator circuitry. In the latter case the chip and PCB temperatures were stably 2ºC lower that is why we would recommend this particular installation as preferable (see the corresponding photograph in our review).
Now it is time to check out the coolers performance with a GeForce 8800 GTX based graphics card, where our today’s newcomers from XIGMATEK and AURAS should face an even more powerful opponent - Arctic Cooling Accelero Xtreme 8800 that impressed us with its extreme efficiency but disappointed a little with lack of universality:
And again both new coolers perform great. Each of them is either as efficient as Arctic Cooling Accelero Xtreme 8800, or outperforms it. We were extremely pleased with AURAS Fridge that provided not only the best GPU cooling, but also the best PCB cooling over the entire test session.
Now let’s check out the coolers’ noise:
Since we tested AURAS Fridge with the same fans as XIGMATEK cooler, I decided not to add the results for AURAS with two 92-mm fans to the diagram (they were within the ±0.1dBA measuring error from BATTLE-AXE). Other coolers range from the noisiest one to the quietest. Both new solutions generate very little noise in quiet fan modes and are uncomfortably loud in case of maximum fan rotation speeds. However, AURAS Fridge equipped with a 120x25mm fan working at ~800RPM is practically silent.
Two new graphics card coolers we have discussed today made a very positive overall impression. Nevertheless, I cannot help pointing out a few drawbacks that we noticed. XIGMATEK BATTLE-AXE VD964 has very large distance between the heatpipes in the cooler base, which may theoretically cause local overheating of some graphics chips that do not have a heat-spreader. To be fair I have to say that nothing like that happens on a G92 GPU and the cooler featuring heatpipes direct touch technology demonstrates impressive efficiency. I certainly cannot disregard the fact that the coolers block three PCI slots next to the graphics card and that the new coolers do not remove hot air outside the system case like the contemporary reference coolers do. We can also complain about the fact that XIGMATEK BATTLE-AXE VD964 cooler is missing the fan rotation speed controller and generates quite a bit of noise, while AURAS Fridge comes without any 92mm fans and proper retention for them in the accessories bundle.
You may be overwhelmed with these drawbacks, but the truth is that they all step back when it comes to unprecedented efficiency of these cooling solutions. These two graphics card coolers are indisputable performance leaders on graphics cards with bare GPUs as well as those with heat-spreaders on them. They outperform such well-known opponents as Arctic Cooling Accelero S1 and Accelero Xtreme 8800. Moreover, unlike the latter, XIGMATEK and AURAS solutions are universal coolers, which means that they make a much smarter investment than Xtreme 8800/2900.
All in all, I can conclude that today overclockers got another two highly efficient graphics card cooling solutions. And those who haven’t yet made up their mind to replace the roaring reference cooler on their hot graphics card got another reason to think things over one more time.