Today we will be looking at the Arctic Accelero Xtreme Plus heatsink (try saying that name five times fast!) Arctic, previously known as Arctic Cooling, is a brand name that most computer hardware enthusiasts will recognize for their large selection of products ranging from MX-2 thermal paste and Freezer-series CPU coolers to, of course, Accelero graphics card heatsinks. Their heatsinks have a reputation for providing quiet, high performance cooling at a reasonable price.
Over the past few weeks I have used this cooler on my GTX 580 (1.5 GB) to test its performance and compare it against the following heatsinks: Thermalright Shaman, Asus DirectCuII, MSI TwinFrozrIII, and the stock rear-exhaust reference cooler.
Specifications and Features
(Courtesy of Arctic)
- Dimensions: 290 mm x 104 mm x 56 mm (L x W x H)
- Net Weight: 615 g
- Heat Pipes: 5 x 6 mm
- Base: Copper
- Fans: 3 x 92 mm (900 – 2000 rpm) – Controllable by PWM
- Airflow: 81 CFM
- Limited Warranty: 6 Years
- PCI Slots Used: 3
- Patented fan holder to reduce buzzing sounds
- Pre-applied Arctic MX-4
- Crossfire, SLI compatible
- Maximum Cooling Capacity: 250 Watts
- AMD: 6970, 6950, 6870, 6790, 5870, 5830, 4890, 4870, 4850, 4830, 3870, 3850
- nVidia: 580, 570, 560Ti, 560, 550Ti, 480, 470, 465, 460 SE, 460, 285, 280, 275, 260+, 260, 250, 9800 GTX+, 9800 GTX
The packaging for the Accelero Xtreme Plus (AXP) is a fitted clear plastic clamshell case.
The back of the packaging contains the list of features, specifications, and compatibility information (including the list of accessory kits to match your video card).
Inside the packaging we find the cooler, a VGA bracket, installation instructions, screws and washers, and an ATX adapter to control the fan speeds set at 7 V or 12 V (if you decide to not use PWM).
The installation of the AXP really couldn’t have been any simpler. Honestly, removing the stock heatsink from my GTX 580 was more work than attaching this cooler to it. After removing the reference cooler from my GTX 580 I used some canned air, a paper towel, and some rubbing alcohol to get rid of any of the dust that had collected. Once the card was dry, I proceeded to clean the VRAM and VRMs using a standard pencil eraser (as recommended in the included instructions). After all of the modules were cleaned, I opened the thermal adhesive that came with the VR004 accessory kit and began to apply it to each of the VRAM chips and VRM MOSFETs. I had to be careful not to over- or under-apply the glue as the tube of adhesive was relatively small, but there was still some left in the tube after I was done so it shouldn’t be an issue unless you plan to do multiple installations. I then installed the VR004 set of heatsinks to the card. It was getting late when I finished so I left the card to cure overnight (the instructions advise a minimum time of one hour).
When I came back to the card the next morning, all of the heatsinks were adhered in place so I proceeded to attach the mounting plate and then installed the video card onto the cooler with the included screws and washers. For consistency between each of the tests, I cleaned off the MX-4 from the base and applied MX-2 thermal paste; the paste was applied by spreading a thin layer over the GPU core with a razor blade.
Once the cooler was attached, I installed the video card into my motherboard and attached the 12 V ATX adapter in order to observe the maximum fan speed/noise, then turned on my PC.
Test Setup and Methodology
- Intel Core i7 2600k
- Asus P67 Maximus IV Extreme
- 2 x 2 GB G.Skill Ripjaws DDR3-2200
- Zotac GTX 580 (1.5 GB) – Modded BIOS for up to 1.212 V
- Arctic MX-2 Thermal Paste
- nVidia ForceWare 285.38 Drivers
- Silverstone Gold 1250 W PSU
- Navig Benching Station
- Ambient Temperatures = 24 °C (+/- 0.5 °C)
The methodology used for this review was as follows: The various coolers were set to their maximum fan speed and the GPU Vcore was increased by 0.025 V between each test, starting at the stock 0.975 V to the BIOS-modded 1.212 V, while running the GTX 580 at its stock speeds (772 MHz core clock, 1002 MHz memory frequency). The GPU was put under load by running Unigine Heaven 2.5 with the Extreme preset at a resolution of 1920 x 1200, with all of the settings at their maximum (Extreme Tessellation, High Shaders, 16x Anistrophic Filtering, 8x Anti Aliasing). The maximum temperatures were measured after a full pass of the benchmark with MSI Afterburner. I gave the card 3 minutes of downtime to allow the card to cool down between the tests before changing the voltage and starting the next test.
I was assisted in my review by one of my fellow forum members: MattNo5ss; he tested the MSI TwinFrozrIII using the GTX 580 Lightning, Thermalright Shaman, and Asus DCuII using the GTX 580 Matrix. The heatsinks were tested using the same parameters that are listed above to keep everything constant. He also tested the sound levels of the various heatsink/fans: He manually set the fans to run at 100% (in the case of the AXP it was set directly to 12 V) then measured the sound using a sound level meter from two feet away and recorded the dBA levels.
Here’s a picture of the AXP with a test in progress.
Here are some pictures of the AXP’s competition.
The first graph shows how the coolers fared when put under stress with the intense graphical benchmark Unigine Heaven. You may notice that there are no results for the DCuII until the 1.050 mark; we were unable to have the Asus Matrix card complete a full run with less voltage, so we used that point as the starting marker for that cooler.
As you can see in the graph above, the Stock/Reference cooler was the worst performer of the collection throughout each of the voltage levels, hitting a maximum of 67 °C at the maximum voltage (1.213 V) that we applied. The AXP fell right into the middle of our test subjects, reaching a highest temperature of 60 °C during a full pass of Heaven. The Asus DirectCuII cooler is the overall winner in terms of total cooling capacity with a maximum temperature of 54 °C. While those numbers are excellent for the DirectCuII (compared to the 4 other coolers), let’s see how the various coolers did when we take noise levels into account.
As a reminder, we measured each of the heatsinks after they were set to 100% fan speed, and we used a sound level meter to read the dBA levels of the fans while we were running the benchmarks.
As you may notice, the stock HSF is not listed; unfortunately we did not have a sound meter to check the numbers of the stock fan prior to the other HSF installations and measurements. Upon searching for results from other reviews we were unable to find a list of results that met how we did our testing (100% fan speed measured from 2 feet away). With that said though, from listening to the DirectCuII fan most recently and comparing to the noise of the stock fan, I would estimate it to be as loud or louder than the rest of the heatsinks at 100% fan speed. Also, for a comparison, with both AXP and stock fans running directly next to each other, I found that the AXP running at 100% (12 V) was approximately equal in noise to the stock fan running at 40-45% speed (set in MSI Afterburner).
These results show us that unless you are hard of hearing or wear noise-cancelling headphones, the Twin Frozr III, DirectCuII, (and from my personal experience, the stock heatsink) produce a lot of noise while running at 100% fan speed. While the louder fans may create a difficult environment to hold a conversation with normal speaking voices, the AXP and Thermalright Shaman are going to be among the quietest parts inside your computer, even running at their maximum speeds.
For one last test, we decided to match up the top-cooling Asus DirectCuII against both the Accelero Xtreme Plus and the Shaman to see which cooler wins out with low noise. We measured the sound output of the DCuII from 2 feet away at various fan speeds and found that at 43% the noise production reached 36 dBA, the same as a full speed AXP and then put it through the test series one last time.
As you can see, the DCuII cooler still performs quite well, but it doesn’t quite reach the performance levels of the Accelero Xtreme Plus cooler, and the Shaman still comes out on top in terms of both cooling and noise production.
As we can see from the results, Arctic’s Accelero Xtreme Plus performed quite well compared to the competition, cooling between 6 -10% better than the stock reference heatsink at all voltage levels. The coolers that come stock on the Asus DirectCuII (and Matrix) line cooled the best of those that were tested, although it was at a significantly higher sound level compared to the Accelero and the Shaman. Overall, the Shaman beat out the Accelero Xtreme Plus in both cooling and (lack of) noise production, but according to Thermalright’s spec sheet it will take up 4 PCI-E slots on a motherboard. This will make it difficult to have a SLI setup depending on the spacing of the slots, unless the Shaman was only used on a single card in the lower of the two slots (possibly impacting other expansion card slots). The Accelero Xtreme Plus (I) takes up 3 slots (akin to the DirectCuII) and has an incompatibility with cards (or at least GTX 580) that have a stacked DVI port design which is not uncommon in non-reference models — if you have a stacked DVI solution you will need the Accelero Xtreme Plus II. While the stock and TwinFrozrIII cards take up two slots, they were also the lowest performing solutions of the bunch and produced the most noise. With the potential spacing issues for a SLI or CrossFireX solution, I must recommend to check the spacing between your motherboard slots before purchasing the AXP. For overall value, the AXP costs between $65 and $75 at various e-retailers, but you must also factor in the cost of the accessories kit (Socket/Base adapter, RAM sinks and VRM sinks) which are ~$8 directly from Arctic. Overall, for a product that gives better cooling than the stock fan at a whisper-quiet noise level, this heatsink earns our Overclockers Stamp of Approval.
– Don Fisher (Janus67)