Star Rating

Performance (PWM-Controlled)

We begin our benchmarks with PWM-controlled results, using our Gigabyte motherboard's default PWM fan profile, which increases fan speed as temperature sensors detect increasing heat. Note that the actual RPM is not controlled by the motherboard, only the voltage, so fans with top speeds of 1000RPM, 1500RPM, or 2000RPM, for example, will behave very differently when receiving the same PWM signal.

First up, idle numbers. Note that while our processor is overclocked, it still enters an idle state, reducing voltage substantially.

Idle Performance

We want to be very clear here. If you're focusing on temperature reading in the graph above, you're doing it all wrong. At idle, the only thing that matters is noise. Attempting to achieve something meaningful by having the coolest-running PC at idle is a fool's errand! The co-winners here, by the slightest of margins, are the Reeven Ouranos and Thermalright Macho Rev. B, the former by virtue of its excellent fan, and the latter by virtue of its slow fan. Not that it matters, but the Macho's immense heatsink is so effective at low fan speeds (and even at zero fan speeds, if you dare!) that it has by far the lowest idle temperature.

Next up, the multi-core benchmark built into the excellent (and free!) system utility, CPU-Z. While we used to use a gaming-related benchmark as our second test, we found that the fan noise generated by our video cards under such conditions really threw off the results. As it happens, CPU-Z generates a load very similar to a modern multi-threaded game engine, so if you're a gamer, this is the result you want to look at.

CPUz Performance

And here, things start to get really interesting. First, note that our least-expensive and least-capable cooler, the AR07, out-performs all other models when used with a stock 6700K. This is critically important for our readers to understand: there simply is no way to overheat a 6700K with a halfway-decent cooler if you're not overclocking. It's just not going to happen, so don't sweat the details if you're not an overclocker... just cap your CPU cooler budget at $40 and never look back.

As for our overclocked numbers, well, the AR07 falls way back, joined oddly enough by the most expensive cooler in our test, the downdraft-style Noctua NH-C14S. This is going to be a continuing pattern, and one that's very important to keep in mind if you have room for a tower cooler. Despite the fact that the NH-U14S and NH-C14S are essentially identical in everything but orientation, they couldn't be more different in performance.  That being said, even the NH-U14S isn't at its best here, as two less expensive but bulkier coolers are able to out-muscle it, with Thermalright's Macho achieving the enviable goal of lowest temperature and lowest noise level.

Finally, we illustrate cooling performance faced with the insane load generated by Intel Burn Test. We don't think any application will put this much pressure on a CPU, so consider this an absolute worst-case scenario.

IBT Performance

The good news: all of our coolers easily passed 20 cycles of this test without even coming close to the 100 °C thermal throttling threshold of the Core i7-6700K. Again, at stock clocks, the AR07 is more that a match for this CPU, with a lower temperature than our other contenders can achieve once the CPU is overclocked. Once overclock, our mid-priced models continue to impress, although the Ouranos is starting to sound a bit rough, owing to its high-RPM fan. We're a bit surprised the NH-U14S didn't do better in these PWM tests, but as we'll see on the next page, it has a different trick up its sleeve.

Previous page Next page