In April I compared four fanless coolers to see how they performed relative to each other. Since then, the SilverStone HE02 cooler has been discontinued, and I wanted to find a suitable replacement for it. So, I decided to test an additional five large air coolers without their fans to see if I could find one that performs as well as the HE02 did in a passive configuration. Included in this round of testing are the Scythe Ninja 5, Raijintek Tisis, Thermalright Le Grand Macho, Phanteks TC14PE, and Prolimatech Megahalems.
Looking back, the results from my last round of tests were not very accurate. I believe that the ASRock H370 Pro4 motherboard that I used was underreporting temperatures and/or overreporting power use. Temperatures were about 30% lower than I would normally expect for the coolers tested, based on what I’ve seen with other motherboards. That said, because I used the same setup for each cooler, the results could still be useful for comparative purposes. I included the NoFan CR-95C cooler in this second round of testing to see how the previous results could be scaled to match the results from this round.
I tested the six coolers in two system configurations at various power levels to see how each cooler does under the same sets of conditions. The first configuration included the Intel i7-10700 on an Asus Z490-Creator motherboard. The second included the AMD Ryzen 7 Pro 4750G on an Asus B550-F motherboard. The Fractal Design Meshify C case was used, and was stripped down, all fans, drive bays, and dust filters removed for optimal natural passive airflow. I ran Prime95’s torture test with small FFTs for half an hour during each run. This was done a total of 88 times, running each CPU under each cooler at 10-watt power limit intervals from 10 watts up to the highest power level that each cooler could handle without CPU throttling. I changed CPU power limits quickly between each test so that the coolers lost as little heat as possible. This way I could get to heat saturation more quickly during each run. CPU core temperatures were tracked with HWINFO64, saved as CSV files, and then processed in Excel. I ended up with a summary of average core temperatures for each minute of each stress test.
For simplicity, I’m only including Intel in this first set of graphs. Here are the results for the Scythe Ninja 5. The y axis here is average core temperature in Celsius above ambient. The ambient room temperature was subtracted from the average core temperature for each minute the tests were run. The Ninja 5 passed the 70-watt test without any throttling, but not the 80-watt test. Maximum temperatures increased by an average of about 8 degrees each time the power limit was increased by 10 watts.
These are the results for the Raijintek Tisis. The Tisis also passed the 70-watt test, but not the 80-watt test. Maximum temperatures also increased an average of 8 degrees for every 10 watts of power use.
The Thermalright Le Grand Macho was the first cooler to pass the 80-watt test. Maximum temperatures increased an average of 8 degrees for every 10 watts.
The Phanteks TC14PE also passed the 80-watt test, and maximum temperatures again increased an average of 8 degrees for every 10 watts.
So far, each change in coolers brought lower temperatures. Here’s where that changed. The Prolimatech Megahalems barely passed the 70-watt test, and maximum temperatures increased an average of 9 degrees for every 10 watts.
Finally, the NoFan CR-95C yet again proved itself as the leader of the pack. It passed the 90-watt test, and temperatures increased an average of only 7 degrees for every 10 watts.
Here is a summary of results with the Intel i7-10700. Power is on the x axis, and core temperatures above ambient on the y axis. Linear trendlines were added, and formulas were determined for each cooler. All of the coolers here were somewhat comparable in performance, although the NoFan CR-95C was clearly the best performer, and the Prolimatech Megahalems clearly the worst.
Here are the results with the AMD Ryzen 7 Pro 4750G. The AMD tests ended at 65 watts because the 4750G would not allow me to increase power limits above that point. Results between the coolers were a little more indistinguishable, but other than the NoFan cooler, the Phanteks TC14PE also stood out a bit ahead. The smaller Prolimatech cooler again fell at the back of the pack. The Raijintek and Thermalright coolers did better than the Scythe cooler with Intel, but here they both fell a little behind the Scythe cooler. I believe this was due to their orientation in the case. The Scythe cooler has radial symmetry, so it doesn’t matter in what direction it is installed. However, both the Raijintek and Thermalright coolers could not be installed in their optimal orientations for passive cooling with the AMD CPU because of the lack of radial symmetry with the AM4 mounting holes.
If we average the Intel i7 and AMD Ryzen 7 results this is what we get. The differences between the Intel and AMD sockets balanced each other out, and the Scythe Ninja 5, Raijintek Tisis, and Thermalright Le Grand Macho cooler average performances were almost identical. The Prolimatech Megahalems is a little behind still, the Phanteks TC14PE a little ahead, and the NoFan CR-95C significantly ahead. I haven’t tested the first five coolers before, but I have tested the NoFan cooler quite a bit, and these results line up well with what I would expect based on my experience with the cooler. So, unlike with my first fanless cooler comparison, I have a lot of confidence that these results do a good job representing what you should expect if used in a case that has a lot of natural ventilation.
For each cooler here I used the trendlines to estimate at what power limit I would expect to see core temperatures hit 100 degrees when fully stressed, assuming an ambient room temperature of 25 degrees. For the NoFan cooler this gave me a power limit of 95 watts; exactly what the cooler is marketed for! I’ve added the scaled down results from my previous tests for the SilverStone HE02, NoFan CR-80EH, and Arctic Alpine 12 passive coolers. The five new coolers tested all happen to fit in between the NoFan CR-80EH and SilverStone HE02 in performance.
Now let’s look now at a broader comparison of these coolers. The theoretical maximum power was calculated as just mentioned, again assuming an ambient 25 degree room temperature. The thermal resistance represents the expected increase in CPU temperature for each additional watt. The Megahalems and CR-95C coolers do not support the AMD AM4 socket, while the remaining four do. I had to do some modifications to these coolers to be able to test them with the Ryzen CPU. All of them support the Intel LGA 1200 socket. The cooler heights are mostly similar, with the Raijintek being the tallest, and the NoFan being the shortest. Mass is also mostly similar, although the Prolimatech and NoFan coolers are a bit lighter than the others. Color options, installation difficulty, and current USA prices are also listed for each cooler. Considering that performance was similar for most of these coolers, the Raijintek Tisis looks like the best value, at $53 at Newegg. This price is for the black Core Edition, which does not include fans. For the better performers, the Phanteks cooler is available for about $90, and the NoFan cooler is still available from QuietPC USA for $135, including shipping. A few additional notes were added at the bottom. The Thermalright cooler’s asymmetric design works well in an upward configuration, which is possible on the Intel LGA 1200 socket, but not on the AMD AM4 socket; and the NoFan CR-95C will block 1 or 2 PCIe slots on the motherboard and does not support Mini-ITX motherboards.
If you are considering a fanless CPU cooler to help silence a standard tower style PC system, I hope these results are useful. Keep in mind that the case chosen and room temperature will have a significant effect on CPU temperatures. I am eagerly awaiting Noctua’s pending passive CPU cooler, which is on their roadmap for release in the first quarter of 2021. Let me know if there is another cooler that you would like to see me test passively in a third passive cooler showdown.