Performance vs. the Stock Intel Cooler
Let's preface the performance results by saying that our expectation is that both the Noctua NH-L9i and SilverStone AR11 should out-perform the Intel cooler. If they didn't, there'd be absolutely no reason for their existence (well, except perhaps to fill in for all the Intel coolers missing from "K"-series CPUs that Intel currently sells). And to make a finer point of what we mean by performance, we're talking about both temperature levels and perhaps more importantly, noise levels. Our rationale is that the Intel cooler is already guaranteed to function within acceptable temperature limits on a stock-clocked processor. If it didn't, Intel would include it (and frankly, this may be why it's not included with "K"-series processors anymore - it's wholly unsuitable for overclocking).
With that said, we're going to ruin the surprise a bit. The SilverStone AR11 cooler is too loud. We realized this right away, and for that reason, we've added a "fourth" cooler to our graphs: the AR11 with the Noctua NF-A9x14 fan from the NH-L9i cooler, which was a direct fit. Pay close attention to how the Noctua fan changes the behavior of the AR11 heatsink in the graphs below.
We start with idle performance, where it's all about noise, because heat is simply not an issue in this mode. Here, Noctua proves it knows a thing or two about cooler design, handily outperforming the Intel cooler while staying quieter. The SilverStone AR11 achieves enviable temperature levels, but it's all for nought, as the noise level is insanely high given that heat is simply not an issue here. You can thank SilverStone's overly-conservative minimum fan speed of 1160RPM (officially rated at 1200RPM, by the way). So what about the AR11 fitted with the slower-turning, higher-quality Noctua fan? Well, it's nearly as cool as the stock AR11, but shockingly not as quiet as the NH-L9i. What could the reason be, given that the fan is the only noise-producing component of these CPU coolers? Well, it would take a bit of detective work to find out, so stay tuned for that on the next page.
When running CPU-z's built-in benchmark, which closely approximates the load created by typical application use, the Pentium G4620 stays safely within its thermal limits on all of these coolers, even the Intel cooler, which is perfectly capable even at its minimum RPM. But what's most interesting is how the Noctua, SilverStone, and Noctua/SilverStone hybrid compare. As you'll see, the SilverStone continues buzzing away while hitting extremely-low temperatures, but when fitted with the Noctua fan, it gives up a bit of that performance while dropping substantially in noise. This is a very fair tradeoff from our point of view, but it begs another question: is the AR11's heatsink in fact better than the Noctua's?
We think the results speak for themselves. When all else is equal (i.e., the fan and fan speed), the AR11 just barely squeaks by the NH-L9i. This may not be surprising, given that it stands 10mm taller, but remember, it's also over 100g lighter. Based on these results (which are further supported in the graph below), it appears that height does indeed trump heft when it comes to air cooling. In other words, it takes more than a block of heavy metal to cool a CPU!
OK, time for another test, this one focused on mathematical calculations. The TimeSpy CPU test, while part of a 3D gaming test suite, is actually all about the CPU, and here, it's clearly applying a heavior load to the CPU than CPU-z's benchmark does. The coolers all fall in line as expected, with the Intel cooler being competent but not convincing, and the SilverStone cooler being greatly over-powered for this application. Note again, however, how the AR11 using the Noctua fan is a bit louder than the Noctua NH-L9i, while only a bit cooler. It's nearly as loud as the Intel stock fan.
More testing is in order, and we were going to have to get a whole lot more creative to figure out what was going on!