Video Discription |
https://www.patreon.com/fullysilentpcs
https://www.fullysilentpcs.com
Two and a half years ago I compared a total of nine CPU coolers without any fans to see how they performed in a fanless PC build. The NoFan CR-95 and SilverStone HE02, two coolers designed for fanless cooling, performed the best, but several other large tower coolers performed quite well too, even without the fans that came with them. Unfortunately, most fanless coolers haven’t had a lot of commercial success, and most of the coolers in that comparison have been discontinued. Luckily, June of 2021 saw the release of Noctua’s NH-P1, which, weighing over a kilogram with 13 extra thick heatsink fins, delivered excellent performance, beyond any fanless CPU cooler we’ve seen. How much better is it really though? That’s one question I plan to answer here. I also managed to get my hands on another four coolers to test along with the P1, the Zalman FX70 and FX100, the Prolimatech Genesis, and the Raijintek Ereboss. Three of these have been already discontinued for years, but I still included them out of curiosity and because the results would get us closer to understanding what design features make for a high-performing fanless cooler.
I added the Arctic Alpine 12 Passive, Prolimatech Megahalems, and NoFan CR-80EH and CR-95C for comparison and scaling with previous comparisons. I tested the nine coolers with two CPUs, the Intel i9-11900 and i9-12900, at 15-watt power limit intervals. Each test included a half hour Prime95 stress test, during which I used HWINFO64 to record CPU core temperatures. There were a total of 74 test runs, and the data was processed in Excel to create a temperature vs power limit trendline for each cooler. Creating these trendlines allows me to verify the quality of the data, and estimate the expected maximum temperature of other CPUs at any power level, with each cooler. Each data point represents the average core temperature right at the end of each half-hour stress test. The Fractal Design Meshify 2 Compact case was used with all fans and dust filters removed for optimal natural passive airflow. For the i9-11900, an Asus Strix Z590-A Gaming WiFi motherboard was used, and for the i9-12900, an Asus Prime Z690-A motherboard was used.
Here are the results, all on one page. The X axis is the power limit in watts and the Y axis is the maximum average core temperature in Celsius. So, the lines at the bottom and right side of the graph performed the best, and the lines at the top and left side performed the worst. The results from the 11900 and 12900 were averaged for each data point here. The core temperatures were normalized for a 25-degree ambient temperature. Unexpectedly, the budget friendly Alpine 12 Passive performed the worst. Also unexpectedly, the NH-P1 performed the best. What did surprise me was how well the two long-discontinued Zalman coolers performed, especially the FX70. At less then half the weight, the FX70 was actually competitive with the Noctua P1. Zalman’s unique twisted heatink fin design seems to have worked extremely well. The FX100 performed very similarly to the CR-95C, the Genesis fell significantly behind the FX100 and CR-95C, and the Ereboss fell a bit behind the Genesis. They still did better than the CR-80EH and Megahalems though.
You might notice the formulas below each cooler name in the legend. These represent each linear trendline. You can use these formulas to calculate the theoretical maximum core temperature at any power level for each cooler. They way I like to compare coolers is with the power limit at which I would expect to see thermal throttling begin. If we use Intel’s Tjuction of 100 degrees for the y variable of each formula, the x variable will be the thermal throttling power limit.
Here is what we get for each cooler. They don’t exactly match the formulas on the previous graph, because I averaged those results with results from previous comparisons. If top performance is what you are after, the Noctua NH-P1 is your best choice, at 116 watts. For its lighter weight and lower cost though, while it was available, I think the Zalman FX70 had the best overall design, and handled 105 watts. In 2013 the Zalman FX100 was marketed as the ultimate fanless CPU cooler, but with 89 watts, it fell a bit behind the NoFan CR-95C, which had already existed at that time, and the FX100 was later outdone by Zalman’s own FX70. If you are looking for a budget option for a somewhat lower-power CPU, the Raijintek Ereboss does a great job with 74 watts. Finally, of the newly added coolers here, the Prolimatech Genesis did surprisingly well at 79 watts, but I’m not sure its odd design would justify it as an optimal choice for fanless use.
If you are considering a fanless CPU cooler to help silence a standard tower style PC, I hope these results are useful. Keep in mind that the case chosen and room temperature will have a significant effect on CPU temperatures. [FE56m0Vgs04] |