One of the most popular CPU coolers of today has already appeared in our tests, but on an older, basically outdated platform. That’s why we’re bringing you measurements that better reflect what’s in use these days. We have the results of the Endorfy Fera 5 cooler on the latest Intel and AMD platforms with their top processor models. How does an, essentially, inexpensive tower cooler handle them?
Conclusion
Among inexpensive models, an excellent cooler for the middle class of modern Intel and AMD processors. By this we mean Core Ultra 5 (Intel), possibly Core i5 or Ryzen 5 9000/7000 (AMD) processors.
Based on our tests with more powerful CPUs, we can say that the CU9 285K without power limits was always beyond the capabilities of the Endorfy Fera 5 (Black) cooler. Even at the maximum speed of its fan, when the highest cooling performance is achieved, the P-core clock speeds were about 400 MHz below what we measured with 360 mm liquid coolers (Navis F360 and Lumen S36 V2). On top of that, this was the case with the cores (CPUs) being heated to over a hundred degrees Celsius, which naturally already hit thermal limits significantly.
We’ve managed to tune the Endorfy Fera 5 Black cooler to the quietest mode, but again – the available cooling performance is for lower-power CPUs. In this particular case, at very quiet operation, that’s less than 125 W for the Intel platform. That means below the TDP level of the “K” models. You probably won’t reach even base clock speeds in such conditions, but for quiet operation of 65-watt models, the Fera 5 Black cooler will be suitable. And there may even be talk of some performance headroom.
From the perspective of AMD processors, the situation is more favorable even at very low speeds, at which, although this cooler will not be able to maintain the maximum performance of the Ryzen 9 9950X, it is not far from being able to do so. At medium speeds, the cooling performance of the Fera 5 (Black) is already just right even for such a powerful processor. Of course it is also true that it is an optimal cooler for lower-end processors on the AMD platform. Although there’s the option of working with the lower-power Ryzen 7s, keep in mind that these processors only have one CPU core chiplet active. In other words, their heat flux can be denser than that of two-chiplet models with higher overall power consumption. In such cases, the heat dissipation requirements are even more stringent.
Although compared to the liquid coolers tested so far, the Fera 5 Black is weaker, it should also be noted that its price is 4–5 times lower. The price to cooling performance ratio of this cooler is really attractive. Nevertheless, the Fera 5 variant with a black heatsink costs extra compared to the “standard” (with a heatsink without the black coating). The latter is cheaper and has an even better price/cooling performance ratio. In any case, the choice of Fera 5 Black will be attractive for those users who also care about the appearance or matching the colour scheme of the entire build. Here, the black design often fits in better than that which is natural for an aluminium heatsink.
In its price range, the Endorfy Fera 5 Black has an extremely high cooling efficiency on today’s processors, for which it deserves the “Smart buy!” editorial award.
English translation and edit by Jozef Dudáš
| Endorfy Fera 5 Black |
| + High cooling performance... |
| + ... also suitable for more powerful processors |
| + Design works well with both mainstream platforms, Intel and AMD |
| + Wide fan speed range... |
| + ... with optimization options for very quiet but also high performance operation |
| + Very attractive price/cooling performance ratio |
| + Visually appealing all-black design |
| + Easy and quick installation |
| - It may not be able to handle the most powerful processors, but this cooler was not built for those |
| Approximate retail price: 28 EUR |
Is the Arctic Freezer 36 the number one challenger? 🙂
I really like this methodology! Performance is the metric we should be comparing first and foremost, not temperatures. Also, I highly appreciate having noise and frequency analysis on the pumps only.
Some observations:
1. If processor is NOT thermal throttling: Faster fan speed = lower temperature, equal or higher clock speeds, lower power consumption.
2. If processor is thermal throttling: Faster fan speed = same temperature (Tjmax), higher clock speeds, higher power consumption.
3. For AMD, lower temp does not equal to higher frequency, for example Fractal Design Lumen S36 v2 always achieve higher clock speed than Endorfy Navis F360 on 9950X. Some investigation is needed – it’s perhaps due to different cores having different temperatures. In that case, some extra data might need to be included (e.g. temp of the hottest core).
4. Although it might have been true that Intel processors are cooler at the same power draw vs AMD, it seems to be no longer the case for this generation.
Suggestions:
1. Sort by processor first – so it goes Intel clock speed, Intel temps, AMD clock speed, AMD temps. I think this is more logical as we need all three data to properly evaluate the cooler’s performance.
2. Frequency is the most important metric, so it should be first, before temperature and power. Temp and power are equally important so their order does not matter, as long as they are next to each other.
3. I think the observations are worth writing an analysis article for, perhaps after more cooler reviews using the new methodology. I don’t think people would be accustomed to the performance-first approach, so the more explanation, the better (perhaps also include interpretation instructions on the result pages?)
Thank you for the useful insights. 🙂
I agree, you’re right. CPU core clock speed is an important metric and could be listed first among the graphs. We originally had it that way, but in the end, the temperature is what’s on top. The reason for this is that people are used to comparing this parameter and are more comfortable with it when comparing coolers. But this parameter can be misleading and that’s why for completeness we also list the clock speed in MHz and the power consumption in W. We will do our best to educate the readers gradually and… we will see.
Also noteworthy is the comparison of the shape of the consumption curve, which is different for coolers that do not have performance difficulties (e.g. Navis F360) than for “problematic” ones (e.g. Fera 5), which cannot sustain a given load. We also plan to publish and analyze these things later in separate thematic articles. Addressing this in standard tests proved impossible from a time perspective. And it would probably be ineffective – such an analysis would get lost among other things and fewer people would get to it.
Yes, it would certainly be useful to monitor and record the clock speed of each core separately. This will of course vary depending on how much contact is there with each cooler over a particular core. When the core clock speeds are averaged, as we do, what you describe can happen. I’m wondering how to reasonably interpret this data with respect to sufficient clarity. We’ll figure something out in the future. We record this data in the HWiNFO logs, so we have it available for all the coolers we tested, even the first ones (Navis F360, Lumen S36 V2 and Fera 5 Black).