We have here a GPU cooler here which has the “world’s first” moniker attached to its resumé, the Thermalright Shaman. The eight heatpipes justify this moniker, and it has at least two more heatpipes than any other GPU air cooling solution out there. Over the past couple of weeks I’ve used my GTX 460 1 GB to test the Shaman against Prolimatech’s MK-13 and eVGA’s “External Exhaust” stock cooler. Today we take a closer look at the the heatsink itself and its performance.
Specifications & Features
- Dimensions: 160 mm x 132 mm x 38 mm (L x W x H)
- Weight: 500 g (Excluding fan and mounting brackets)
- Heatpipes: 8 x 6 mm heatpipes
- Base: C1100 pure copper, nickel plated with mirror shine
- World’s First VGA cooler that supports compatibility with a 140 mm fan, and has a dissipating area of 140 mm x 130 mm.
- World’s First VGA cooler with eight 6 mm heatpipes, easily cools the hottest VGA in the market.
- All new installation method provides easy mounting and gives better overall compatibility.
- Supports either a 120 mm or 140 mm fan.
- Includes one Thermalright TY-140 PWM fan, ultra low noise at 21 dBA at max RPM (50 cm testing distance)
- Multiple support back-plate to provide the best protection for your GPU.
- Mirrored copper base with quality craftsmanship.
The Shaman arrived in packaging typical of Thermalright: a plain box with just their logo and the name of the contents on the side. When you have the reputation that Thermalright does, there’s no need for anything flashy on the packaging.
Inside the box, the first things you see are the installation instructions, contents list, and a Thermalright sticker. With the anticipation building, removing the first thin piece of foam reveals the TY-140 fan.
Removing the TY-140 gives you a small taste of what’s to come. Then, removing the thick foam unveils the Shaman from its hiding spot. The white box houses the included accessories.
Inside the assembly package are two plastic bags: one holds the mounting hardware and the other holds the RAM and VRM heatsinks. There are also four anti-vibration strips, two fan clips, and a small tube of Chill Factor III thermal paste.
Meet the Shaman
Now that our game of hide and seek with the Shaman is over, we can get a good look at Thermalright’s newest member of their GPU cooler lineup. As mentioned in the features, this is the first GPU cooler to have eight heatpipes, and it definitely looks beastly. With the heatpipes placed closely together, every bit of the fins should be holding heat that’s ready to be dissipated. My first impression is that this thing looks like it could do a really good job at keeping my GTX 460 cool.
Looking at the side of the heatsink, it doesn’t look very tall, the fins aren’t very tall either, and it’s kind of angled downward from the bends in the heatpipes to the end of the cooler. Let’s see how they fit eight heatpipes into the Shaman. The heatpipes form a crossing pattern with six heatpipes going one way and two going the other way. This may look odd at first, and you may wonder why they are arranged like that. Here is my best guess: This is most likely due to the extreme curve that would be required on the first few heatpipes if they were all going in the same direction.
Let’s have a look underneath the Shaman and check out the bottom. The first thing I noticed was the offset base, which was required for such a “spread out” heatsink. You’ll also notice in some of the pictures later in this article that the heatsink extends past the PCB of the GPU by around 1.8″ or so. Make sure you have sufficient clearance in your case (width-wise) if you plan on purchasing the Shaman. Now, looking at that base, it seriously is a mirror finish and Thermalright was perfectly justified in calling it that in the specifications and features.
Remember, just because a base has a mirror finish (smooth) doesn’t mean it’s necessarily flat; there’s a difference between a flat surface and a smooth surface. No need to worry though, the base of the Shaman is smooth and flat, very flat. I couldn’t see any space or light between the base and the rewards card in either direction, parallel or perpendicular to the heatpipes.
Installation went smoothly, except for one minor hiccup. Before mounting the heatsink, I checked to see what kind of RAM sinks I needed to install. I did this by just putting the Shaman on the GTX460 without RAM sinks to check for any clearance issues. It looked like I only needed one of the flat heatsinks, which is good since I would rather use the other ones.
Upon closer inspection it looked like the mounting bracket of the Shaman overlapped the RAM chips along the top of the card by ~1-2 mm. This ended up being the case: the bracket isn’t straight, it goes up from the ends a little and then straight across. So, when installing the heatsinks on those RAM chips I left a little room for the mounting bracket. I don’t see this affecting performance since the majority of the RAM chips are covered. From an installation point of view, this could be frustrating for someone who didn’t check closely for clearance. If I would have installed the RAM sinks by aligning them perfectly with the chips, then when I tried to install the Shaman, depending on the pressure used, the bracket would have either sat on top of the RAM sinks or made all those sinks pop off.
After the clearance issue was resolved, I was able to start installing the Shaman. The first step was to put the “double threaded” screws on the Shaman, so I made sure to check the instructions for which holes should to be used for my GPU. These screws were easy to install by hand so no tools should have been required. Unfortunately, one of the mounting holes for the GTX 460 on this sample wasn’t threaded correctly, and I had to use some channel lock pliers for one of the screws. This is most likely just an isolated incident, and I was (un)lucky enough to receive that heatsink!
What I really like about the first step was how easy it made step 2: connecting the GPU to the heatsink. The GPU just has to be laid on top of the Shaman then you can continue with the installation. Next, there are some thick rubber grommets that separate the PCB from the backplate. The backplate slides over the screws, and then the washers go on between the plate and the thumb nuts to prevent metal-on-metal contact. Finally, use the thumb nuts to secure the Shaman to the GPU and you’re done (other than connecting a fan)!
Okay, the Shaman is installed on my GTX 460. The fan is installed like they are typically installed on heatsinks, with clips. The fan clips slide into holes on the ends of the heatsink, and then wrap around the fan and snap into the fan’s mounting holes. One thing to note is that the TY-140 has the same mounting hole positions as a typical 120 mm fan, so these fan clips can be used for any fan with 120 mm mounting holes.
Test Setup & Methodology
- Intel Core i7 920
- Asus Rampage III Extreme
- 3 x 2 GB Mushkin DDR3-2000 (8-8-7-24) RAM
- eVGA GTX 460 1 GB (modded BIOS for up to 1.212 v)
- 36 GB Western Digital Raptor
- SeaSonic X-750 PSU
- Ambient Temp: 25 °C
The methodology used was very similar to that used in my Thermalright Venomous X review. I have compared the Shaman to Prolimatech’s MK-13 and eVGA’s EE (External Exhaust) cooler.
I used voltage to increase the heat load, performed on both a stock clocked and overclocked GTX 460. MSI Afterburner was used for both voltage adjustment and overclocking. The voltage was varied between 0.950 V and 1.2 V – I have a modified BIOS flashed on my GTX 460 to allow the additional voltage. The GPU was placed under load using FurMark v1.8.2 until the core temperature plateaued according to Afterburner’s monitoring graph. If the temps did not plateau then FurMark was stopped: e.g. if temperatures climb into the 80 °C range, then I know that the heatsink just can’t dissipate all of the heat because a small voltage increase wouldn’t cause a 25 °C increase in temperatures.
The MK-13 didn’t come with a fan like the Shaman, so I used two different fans in the testing and acquired temperature readings on both the Shaman and MK-13 using the TY-140 fan included with the Shaman. I did a separate round of testing using a 120 x 38 mm San Ace 109R1212H101 on both coolers. All fans operated at 100% speed, which equated to 74 ft3/minute @ 1300 RPM for the TY-140 and 102.4 ft3/minute @ 2600 RPM for the San Ace H101. The stock EE cooler’s fan also ran at 100%, but I’m not aware of the actual specs on that fan.
Thermal paste was applied by spreading a thin layer over the die with a credit card, a method that was easier and far less time consuming than trying to find the best application pattern for the large, rectangular die of the GTX460 which will create a good spread pattern using only heatsink pressure. I’m used MX-2 in this testing as it is already installed on the MK-13. I’d be glad to use Chill Factor III with the Thermalright Shaman in a follow up post, if there is interest.
So here’s my four test setups, I didn’t include a pic of the stock EE cooler.
I estimated the cooling surface area of the Shaman compared to the MK-13 using the following method: I estimated the total surface area by counting the number of fins and measuring the area of the fins by assuming they are rectangles, which they are not, but this provides a decent approximation. Then, I multiplied that number by two, since both sides of the fins are used, to get the total surface area. When determining actively/passively cooled area, I just took the length of the fan used, 120 mm or 140 mm, and determined how many fins fit within that length, then subtracted that many fins out of the calculation. Not perfect, but a good enough estimation.
One metric that isn’t listed in the graph above is the comparison of actively cooled area between the heatsinks. When the H101 is used the MK-13 has 22% less actively cooled surface area than the Shaman, but when the TY-140 is used the MK-13 only has 1% less actively cooled surface area. You’ll notice soon enough that the actively cooled surface area isn’t directly correlated to the cooling performance. Surface area does play a part in performance, but since % actively cooled area doesn’t closely match the additional % cooling performance, I believe it’s the number of heatpipes that really puts a gap between the two coolers.
Now to the good stuff, and the whole point of a review. The graphs below show my recorded temperatures while the GTX 460 was at its factory clocks, 675/1350/1800, and overclocked to 850/1700/2000 (core/shader/memory).
In the first graph, the TY-140 was used with the aftermarket heatsinks and worked rather well. As expected the stock cooler is there to keep the GPU cool at it’s factory clocks and voltage, which puts the GPU at around 63° C during FurMark loading. Both the MK-13 and Shaman just blow the stock cooler away, keeping the GPU at least 16° C cooler, with one exception. When using 1.2 v and the MK-13, the temp spikes by 18° C, from 63° C to 81° C. This means that the MK-13 + TY-140 is reaching its maximum dissipation in this test at 1.175 v. There were no spikes like that between any other increments, so that’s my reasoning behind the spike. What’s really interesting to me is that the temps while using the Shaman almost scale linearly with voltage.
Now let’s check out how the coolers performed when using a more powerful fan, the San Ace 109R1212H101 or just H101. The MK-13 ended up performing better by an average of ~4° C, I excluded the temp spike in this average since it’s an outlier. The Shaman, on the other hand, performed around 6° C better with the H101. Another thing to note is that the spike in the previous graph for the MK-13’s temps is no longer present while using this fan. This leads me to believe that the TY-140 just wasn’t powerful enough for the MK-13 as the heat load increased. Also, the difference in temps between the MK-13 and Shaman was greater than with the TY-140, so the Shaman seems to like the H101 more than the MK-13. If the H101 worked just as well on both coolers, then the curves would just be shifted down the Y-axis while keeping the temp difference the same as the TY-140.
On to the overclocked results. First, I want to mention that the stock cooler couldn’t handle the overclock, and it wasn’t because of temps. The display drivers would crash or the PC would reset after about a minute of FurMark, so I couldn’t get the plateaued temperatures as I did with the aftermarket solutions. The second important note is that while using these overclocked settings, coupled with the high end of my voltage scale, the temps would not plateau and would continue to rise. When this happened, I stopped the testing once the GPU temp exceeded 90 °C. So, that’s why there seems to be missing data points for some of the higher voltages. Also, notice that the temps with the Shaman are still scaling almost linearly with voltage, even with these overclocked settings.
As before, the data in the first graph was captured while using the TY-140 fan. The shapes of the curves in this graph are a lot like the shapes in the stock clocked TY-140 graph. The difference in temps between the two coolers stayed consistent for a while, then the MK-13 + TY-140 starts to run out of steam in the high heat load department. The Shaman allows for an additional voltage bump over the MK-13 before it too runs into problems, but as you can see the problems weren’t temp related like for the MK-13. The problem while using the Shaman with more than 1.150 v was the display driver resetting on me.
As you’ll see in the next graph, the San Ace H101 really shines with these overclocked settings and high voltage. The H101 brought the temp spike at 1.125 v with the MK-13 down by 14° C, but it still didn’t allow for further voltage increases. Although, the H101 allowed me to complete the entire voltage range with the Shaman. As in the stock clocked results, the temp difference between the two coolers is greater with the H101 than the TY-140, which again shows that the Shaman likes the powerful fan more than the MK-13. Not only that, but the temp difference between stock and overclocked settings using the same fan also increases, which means the Shaman performs even better, in general, during high heat load situations.
The performance of the Shaman is top-notch during both low and high heat load environments. The Shaman keeps its heat dissipation efficiency even during the highest heat loads I produced, and with both low and high speed fans. This can be seen throughout the testing in the nearly linear scaling in temps as voltage is increased. The Shaman should be able to cool the hottest of GPUs with ease.
Installation was just okay. The issue I had with the badly threaded screw hole and the close inspection that was required when mounting the RAM sinks brings installation down to an okay. Everything else about it was great, especially how the GPU lays on top of the Shaman while installing the backplate, washers, and nuts. Also, I want to point out again that the Shaman is wider than typical GPU heatsinks and extends past the PCB by almost 2 inches. So, be sure you have room in your case before making a purchase.
The MSRP on the Shaman is $74.95, while the MK-13 can be had for around $60 on various online stores, $65 with the GTX460 compatibility kit. In my opinion, if you are looking to spend top dollar on a GPU cooler, then the extra $10-15 for the Shaman is definitely worth it. The Shaman can out perform the MK-13 by anywhere between 10° C and 20° C, depending on the heat load and fan used. The Shaman is just a better cooler for the money.
Overall, I would recommend the Shaman to anyone who is in the market to get a top-of-the-line GPU cooling solution without going custom water or the extreme route.