Arctic Cooling Accelero XTREME 4870X2 Review

Not so long ago we reviewed a cooling solution for top of the line Nvidia GPU based graphics accelerators – Arctic Cooling Accelero Xtreme GTX 280. As we pointed out in that article, it is a much more acute matter these days to organize efficient and at the same time quiet cooling of the graphics solutions based on AMD processors, than on Nvidia ones. Radeon HD 4870/4890 graphics cards heat up quite seriously and their stock coolers are incomparably noisier than the reference cooling solutions of GeForce GTX 260/275/280/285. And of course, the dual-processor monster, ATI Radeon HD 4870 X2, is beyond all possible consideration in this case, because its power consumption and consequently heat dissipation are the second highest among contemporary graphics accelerators after GeForce GTX 295.

Of course, a graphics solution like that needs an enhanced cooling system, but since the cooler developers are limited by the relatively small size of the card itself and the close proximity of other system components, they either have to sacrifice the acoustics or to keep the card in very high thermal range all the time. For example, the minimum temperature of the reference Radeon HD 4870 X2 graphics card is 85 °C, but this is very poor consolation, because even if you have a large case with good ventilation, the graphics card hits this temperature mark a few minutes into the game or benchmark. And this is when its fan starts to generate ultimate noise…

The engineers from Swiss Arctic Cooling Company didn’t get discouraged by the complexity of this task. Moreover, after they successfully designed and threw into mass production their Accelero XTREME 4870X2 cooler, they claimed that the new solution was 50(!) °C more efficient than the reference cooler and was working 9 times quieter! The claimed total heat that this cooler can dissipate from the graphics card is around 320 W, which is a pretty unrealistic number even for gigantic CPU coolers. These specifications are hard to imagine or make up that is why we decided to get down to a detailed review of the Arctic Cooling Accelero XTREME 4870X2 cooling solution.

Package and Accessories

New Arctic Cooling solution ships in a molded clear plastic casing shaped exactly like the cooler itself.

This type of packaging not only protects the contents against any possible damages, but also allows the potential buyer to check out the front of the cooler without opening the casing. The back of the package lists all major specifications of the new cooling solution and offers two diagrams comparing the performance of Accelero XTREME 4870X2 against Radeon HD 4870 X2 reference cooler.

Besides the cooler itself, the box also contains two aluminum heatsinks for the voltage regulator components, step-by-step assembly and installation instructions, two adapter cables for additional power, screws and a set of thermal pads.

The heatsinks are only about 10 mm tall and are shaped to match the Radeon HD 4870 X2 PCB components exactly.Obviously, Arctic Cooling decided not to repeat the same mistake again that is why they provided their cooler with good heatsinks for voltage regulator components instead of an aluminum plate, like the one we saw with Accelero Xtreme GTX 280.

The new solution is priced around $63.80 and comes with six-year manufacturer warranty.

Design and Functionality

Accelero XTREME 4870X2 looks very similar to the same cooler for GeForce GTX 260/280 we reviewed before. The new solution measures 295 x 96 x 54 mm and weighs 680 g.

Three 92×12 mm fans in a plastic frame cover the entire front surface of the cooler heatsink.You can easily remove the outer casing with the fans and take a closer look at the cooler heart – its heatsink.

The heatsink is formed of two aluminum plate arrays, each consisting of 0.4 mm plates spaced out at 2 mm distance.Both heatsink arrays sit on copper heatpipes 6 mm in diameter. The cooler uses a total of eight heatpipes. They twist in a very beautiful manner on the way from one array to another.

The heatsink is built in a very curious way. As you know, Radeon HD 4870 X2 uses two graphics processors that is why Accelero XTREME 4870X2 has two copper heat-spreader plates. If we look from the side where the graphics card outs are located, then we will see that the heatpipes from first heat-spreader go through the top part of the cooler heatsink, and the heatpipes from the second heat-spreader go through the lower part of the same heatsink and transfer the heat to the smaller heatsink as well.

The aluminum heatsink plates are pressed against the heatpipes, while in the heat-spreader plates we found traces of soldering, which guarantees the best contact.

All three fans rotate with the same speed that can be adjusted in the interval from 1000 to 2000 RPM using pulse-width modulation method (PWM).

The total airflow from the fans is claimed at 81 CFM and the level of generated noise that Arctic Cooling traditionally measures in Sones shouldn’t exceed 0.5 Sone.

Fluid dynamic bearings used inside the fan rotors should last more than the 6 years promised by the manufacturer warranty although the fans MTBF is not specified among the cooler parameters.

The maximum power consumption of each fan shouldn’t exceed 1.8 W. they are powered via default 4-pin power connector on the graphics card.

The base surface of each Accelero XTREME 4870X2 heat-spreader is covered with a thin layer of Arctic Cooling MX-2 thermal paste, which is one of the best thermal interfaces today.After we installed and removed the cooler, we got the following thermal compound imprints.

I wish there were more thermal paste included than just for one installation, but I am pretty sure that after you see the benchmark results, you won’t feel like removing this cooler from your graphics card 🙂

Installation Tips

There is nothing we can say about the compatibility of the new Accelero XTREME 4870X2 cooler, because it is absolutely clear that it can only fit onto Radeon HD 4870 X2 graphics accelerators as it was specifically developed only for this particular card. HIS Company was kind enough to provide us with a proper graphics card for our today’s tests.

The first thing that you need to do is remove the default cooler from your graphics card by undoing all screws along the perimeter of the card and on the retention bracket.

After that you should install the aluminum heatsinks on the voltage regulator components. After cleaning and degreasing the surfaces place the thermal pads bundled with the card on top of them.

After that install the heatsinks over the pads.If you apply a little pressure to the heatsinks, they will stick pretty securely and won’t come off during the rest of the installation procedure, as it may often happen. Once the heatsink have been installed over the voltage regulator components, place the thermal pads over the memory chips and the bridge chip between the two GPUs.

Then connect two twisted adapter cables…

… and fasten the cooler with screws. However, this is where we experienced a problem. The thing is that there were no retention screws bundled with the cooler that we received for review, because the cooler was intended to be installed using default spring-screws and X-backplates that come with the graphics card. However, during the installation procedure we noticed that the default screws were too short. Take a look (default Radeon HD 4870 X2 screw side by side with the new one).

That is why we had to replace the default screws with longer ones, which we received a little later. As we were told by Arctic Cooling officials, they were already aware of the issue and have decided to include a new screw set in the newly produced Accelero XTREME 4870X2. They also posted up the information on their official website for those end-users who bought the old stock instructing them on how to get new screws free of charge (you have to send a request to freezer@arctic-cooling.com and wait for about two weeks).

Before installing the cooler, do not forget to remove the protective plastic film from the aluminum base plate that will contact the video memory chips through thermal pads. After that you can tighten the screws.

The aluminum plate from the Radeon HD 4870 X2 graphics card default cooler is then installed onto the back of the card, where it also contacts the memory chips via thermal pads. Besides, you also insert three screws through this aluminum plate and they securely hold the heatsinks on the voltage regulator components.

Once we completed the installation, we noticed that the PCB got a little bent.We took a closer look at Accelero XTREME 4870X2 and discovered that both heat-spreader plates and the aluminum plate between them do not go along the same straight line. This must be a production defect that occurred during heatpipes soldering to the base, or the issue with just this particular cooler sample. You can easily fix it by hand by bending slightly the first cooler heat-spreader.

The graphics card with this cooler on it blocks two PCI slots next to it. And even though Arctic Cooling claims that this card can be used in CrossFireX configurations, I am very concerned that its efficiency may not be the same with another cooler like that being dangerously close. Here I also have to add that when we installed Accelero XTREME 4870X2 onto our graphics card, the cooler went 25 mm past the farther side of the already long Radeon HD 4870 X2 card. You should keep it in mind when you decide to buy this cooler or will shop for a system case (there should be at least 295 mm from the back of the case to the HDD chassis). You can download the installation instructions from the official company web-site (PDF-file, 2 MB).

Testbed and Methods

We tested the new cooling system from Arctic Cooling inside a closed system case. Our testbed was identical for all coolers throughout the test session and featured the following configuration:

• Mainboard: ASUS P6T Deluxe (Intel X58 Express), LGA 1366, BIOS 1606;
• Processor: Intel Core i7-920, 2.67GHz, 1.25V, 4 x 256KB L2, 8MB L3 (Bloomfield, C0);
• CPU cooler: Nexus FLC-3000 (900-2500 RPM, PWM);
• CPU thermal interface: Arctic Silver 5;
• Memory: DDR3 PC3-12800 3 x 2 GB OCZ Platinum Low-Voltage Triple Channel (Spec: 1600MHz / 7-7-7-24 / 1.65 V);
• Graphics card: HIS Radeon HD 4870 X2 2×1024 MB, 750/3600 MHz;
• System HDD: Western Digital VelociRaptor (SATA-II, 300 GB storage capacity, 10,000 RPM, 16 MB cache, NCQ) inside Scythe Quiet Drive 3.5” silencer and cooler chassis;
• Backup HDD: Western Digital Caviar Green WD10EADS (SATA-II, 1000 GB, 5400 RPM, 32 MB, NCQ);
• Optical drive: Samsung SH-S183L;
• System case: Antec Twelve Hundred (front panel: two Noiseblocker NB-Multiframe S-Series MF12-S1 fans at 820 RPM and Scythe Gentle Typhoon fan at 840 RPM; back panel: two Scythe SlipStream 120 fans at 840 RPM; top panel: standard 200 mm fan at 400 RPM at the top of the case);
• Control and monitoring panel: Zalman ZM-MFC2;
• Power supply: Zalman ZM1000-HP 1000 W (with a default 140 mm fan).

To increase the total system heat dissipation and make it the testing conditions a little harder for the participating cooling solutions we overclocked our quad-core processor to 3.9 GHz (+46.3%) and increased its Vcore to 1.3125 V (+9.4%) in the mainboard BIOS. The memory voltage was at 1.62 V and its frequency was 1487 MHz (7-7-7-14_1T timings).

The testing programs were installed under Windows Vista Ultimate Edition x86 SP1. We used DirectX libraries (from March 2009) and Catalyst 9.7 drivers. We used 10 runs of FireFly Forest test from the synthetic 3DMark 2006 suite in 1920×1200 resolution to warm up the graphics card. We enabled full-screen antialiasing 4x and anisotropic filtering 16x.

Besides, we additionally used FurMark version 1.7.0 that was launched for 15-minute stability test runs in full-screen mode in 1920×1200 resolution.

Since FurMark is sort of similar to Linpack in terms of the workload it created for the graphics cards, I decided not to subject the borrowed High-End graphics cards to any high risks during this test session that is why in this mode we only ran the tests at maximum fan rotation speed. We used RivaTuner v2.24 (created by A. Nikolaichuk aka Unwinder) to monitor the frequencies and temperatures of our card.

The tests were run at least twice for each tested cooling system. The temperature stabilization period between the two test cycles was about 10-12 minutes. The ambient temperature was checked next to the system case or open testbed with an electronic thermometer with 0.1°C precision that allows monitoring the temperature changes over the past 6 hours. During our test session room temperature was pretty high and stayed around 26.5-27°C, which made the testing conditions a little more serious.

The noise level of each cooler was measured after 1:00AM in a closed room about 20sq.m big using CENTER-321 electronic noise meter. Since we can’t use graphics cards outside the system case, we took the readings from a 10 cm distance.

This will allow to minimize the effect from the noises generated by other system case components. Moreover, we have also lowered the rotation speed of all five 120 mm case fans to 520 RPM and disconnected the backup HDD that was outside the Scythe HDD cooling and silencing unit. In this case the subjectively comfortable noise level was around 40 dBA (marked with red dotted line on the diagrams).

Since we didn’t have any alternative cooling solutions for Radeon HD 4870 X2, we are going to compare the performance of Accelero XTREME 4870X2 cooler against the reference cooler in several operational modes.

Cooling Efficiency and Acoustic Performance

Acoustics

First let’s check out the acoustic readings taken off both cooling solutions in their entire supported range with 5% increment.

The promised 6x advantage over the reference cooler is certainly not here, despite Arctic Cooling’s claims, but the difference is very dramatic anyway!

Accelero XTREME 4870X2 fans work very quietly up until 1100 RPM and after that they generate what I would call moderate noise. The reference Radeon HD 4870 X2 cooler gets noisy at about 30-35% speed and after 35% mark its noise gradually transforms into loud roar. I am not sure what this cooler was initially developed for, but definitely not for home users. Undisputed victory of the Arctic Cooling solution.

Cooling Efficiency

Now let’s check out the cooling efficiency of Accelero XTREME 4870X2. First we will look at both testing participants with fans in automatic mode.

The first thing that we notice is that the temperatures of both graphics processors are the same when cooled by Accelero XTREME 4870X2, while in case of a reference cooler this difference reaches 23°C! This is common for Radeon HD 4870 X2 coolers, because the GPU heatsinks are installed one after another along the direction of the airflow coming from the graphics card fans. As a result, the second GPU heatsink is cooled with already hot air coming off the first GPU heatsink that is why its temperature is way higher. Accelero XTREME 4870X2 doesn’t have this issue, so we see no difference in GPU temperatures, which provides the new cooler with a 20-degree advantage over the competitor.

Three next diodes – GPU memory IO, GPU display IO and Ambient – show almost the same temperatures under both cooling systems. However, the temperature readings taken off the three VRM diodes are quite different, and not in favor of Accelero XTREME 4870X2. Radeon HD 4870 X2 reference cooler cools VRM 18°C better than Accelero XTREME 4870X2, but there are two things to take into account. First, 118°C is quite normal for the VRM of Radeon HD 4870 and similar graphics solutions and no one is scared of this temperature. Second, look at the fans rotation speeds that were used in automatic mode: Accelero’s fans didn’t even move when the temperature increased continuing to run at almost minimal speed of 1010 RPM, while the turbine on Radeon HD 4870 X2 sped up from 1400 to 3090 RPM. Now go back to the acoustics readings chart. I don’t think I need to tell you anything anymore.

The next step (and a pretty logical one) will be to check the graphics card temperature in quiet cooler mode. For this purpose we manually set the fan speed on Accelero XTREME 4870X2 to the minimum of 980 RPM or 26%). The reference Radeon HD 4870 X2 cooler was working in a bit more favorable conditions with its fan rotating at 2290 RPM (35%). Here are the results.

Accelero XTREME 4870X2 at minimal fan rotation speed allowed the GPU temperature to increase by only 1°C compared with the temperature in automatic fan rotation speed mode. All other temperatures remained where they were. However, the reference cooler was nearly dying under the thermal pressure from the powerful graphics accelerator, because only a few minutes into the tests the temperatures rose to critical values and the graphics card thermal protection system started to kick in speeding up the fan to the maximum (as you can clearly see from the typical upsurges on the monitoring chart). In other words, if the reference cooler failed this test even in moderate acoustic mode, then what can we possibly say about the quiet mode?

Since Radeon HD 4870/4870 X2 (and later HD 4890) came out, we have known that their cooling systems were extremely efficient at maximum fan rotation speed and it was very hard to find a worthy alternative to them. However, the today’s hero, Arctic Cooling Accelero XTREME 4870X2, leaves absolutely no chance to this dual-GPU graphics card cooler having yielded non-critically only in one reading (VRM diode). It doesn’t make any sense to compare the noise levels of the two coolers in this test mode: everything is absolutely clear.

To make the results a little more illustrative we summed everything up in a single diagram where we also added the results of Accelero XTREME 4870X2 at average fan rotation speed (1480 RPM). The stricken out numbers indicate the readings taken off the reference Radeon HD 4870 X2 when the test has to be manually terminated.

In my opinion, the most impressive group of results is the first one, where we see the temperature of the hottest GPU of the graphics card. It shows that Accelero XTREME 4870X2 in quiet fan mode cools the graphics processors just as well as the reference cooler at its maximum fan rotation speed. And it is simply impossible to compare the coolers efficiency at the same level of noise, because even at the moderate speed of 2290 RPM the reference cooler couldn’t cope with its task and the card got overheated. This way, Arctic Cooling did keep its word, because its new solution does in fact provide a 50-degree advantage over the graphics card stock cooler. Excellent result!

As for the temperature of other graphics card components, we are a little confused with the VRM readings. Looks like the aluminum heatsinks are not enough to ensure the same cooling efficiency as provided by the reference Radeon HD 4870 X2 cooler. Nevertheless, the difference is not as dramatic as what we saw by GeForce GTX 260. Moreover, these temperatures are pretty common for the VRM of Radeon HD 4870/4890/4870 X2 graphics cards and shouldn’t be considered something out of the ordinary.

Arctic Cooling solution proved more efficient also under higher operational load. Too bad that replacing the graphics card cooler and leveling out the GPU temperatures didn’t have any effect on the card’s overclocking potential: we managed to go only as far as 820/3840 MHz end frequency with both cooling solutions. The temperatures of the overclocked graphics card equipped with Accelero XTREME 4870X2 increased only by 2-3°C on each diode.

And the last thing that we should pay attention to in our today’s review. The thing is that the reference Radeon HD 4870 X2 cooler boasts one advantage: it ousts warm air outside the system case, while the new Accelero XTREME 4870X2 barely does it at all (only part of the airflow from the first fan leaves the system through the grid in the retention bracket). In other words, most of the hot air remains inside the system case. Therefore, we have one logical question: how badly will it affect the CPU thermals? To answer this question we used a new Resident Evil 5 benchmark that runs for about 10 minutes and loads not only the graphics card but also the all Core i7 cores. We launched this test in 1920×1200 resolution with enabled full-screen antialiasing 8x in order to increase the video load (the average fps rate was less than 50). We used RealTemp utility to monitor the CPU temperature. The graphics cards coolers worked in automatic mode. Let’s take a look at the obtained results.

True, the peak CPU temperature turned out 2°C higher when the graphics card was equipped with Accelero XTREME 4870X2 cooler. In my opinion this is very small price for the considerably lower graphics card temperatures and incomparably better acoustics. At the same time, it is important to keep in mind that we ran all tests in a well ventilated system case and in other system cases this difference may be more dramatic. However, I doubt that anyone will decide to install Arctic Cooling Accelero XTREME 4870X2 into cluttered Chinese mid-towers…

Conclusion

Arctic Cooling Accelero XTREME 4870X2 is a truly super-powerful cooler for a super-fast graphics accelerator. It leaves the default Radeon HD 4870 X2 cooling solution absolutely no chances of competing even at minimal fan rotation speed neither in efficiency, nor in acoustics. Its installs simply and intuitively, and the rapidly addressed issue with short retention screws indicates that Arctic Cooling cares a lot about their customers. I can’t find any serious issues with this solution, because there are actually none. Large size? High price? Most likely only theoretical compatibility with CrossFireX configurations? All these would be really far-fetched, because the Radeon HD 4870 X2 owners definitely knew what they were getting themselves into when they purchased a large and hot graphics card like that. That is why they will easily agree to these minor inconveniences for the sake of having Accelero XTREME 4870X2. And I am absolutely sure that the cooler we tested today will prove totally up to their expectations.