There was a time not too long ago when CPUs ran so cool that they barely needed heatsinks, but today's high-performance PCs require users to pay a bit more attention to the cooling of their systems. Luckily, PC users have a nearly unlimited range of fans and heatsinks to choose from to keep their CPUs from overheating. Of course, both Intel and AMD include simple heatsinks with all of their consumer-level CPUs that will work just fine, but if you're a PC enthusiast, you're always after more performance, quieter operation, or maybe just a bit more style. Whatever you're looking for, there's a solution out there for you, and in this article, we'll walk you through the various choices you have, the nuts and bolts (literally) of cooler installation, and the performance you can expect from the various classes of coolers available.
We'll be looking at six coolers in total, each targeting a different niche. This is not a head-to-head showdown so much as a primer on the range of coolers on the market, with a focus not only on how they perform, but also how easy they are to install, how much noise they make, and which applications they're best for. The six contenders are shown below, arranged from smallest to largest, starting clockwise from the lower right with Intel's offering.
We’d like to express a very special thank you to Noctua, which provided us with three of the six coolers we tested and made this round-up possible. The six models are listed below, followed by a table providing their specifications:
- The stock Intel cooler: included at no cost with all consumer-level Intel CPUs, this is by far the most widely-used heatsink in the world.
- The low-profile 92mm downdraft cooler: the Noctua NH-L9x65, designed for small form factor systems. Note that while larger downdraft coolers exist, they are typically based on old designs and are outperformed by similarly-priced tower coolers. We don’t recommend using a downdraft cooler unless the purpose is to fit into a small form factor system.
- The 120mm tower-style cooler: the Noctua NH-U12S, among the very best 120mm coolers available, graced with one of best-engineered fans in the world, the NF-F12.
- The 140mm single-fan cooler: the Thermalright Macho Rev. B, the newest iteration of Thermalright’s premier 140mm tower-style cooler, representing what may be the perfect balance of price and performance for high-end PC builds.
- The 140mm dual-fan tower-style cooler: the Noctua NH-D15, the largest, and perhaps most powerful tower-style cooler on the market. It employs dual heatsink towers plus dual 140mm fans to “max out” what can feasibly be achieved by a “big air” cooler.
- The 240mm liquid cooler: the Corsair Hydro H100i, which really needs no introduction. It’s not the newest liquid cooler anymore, but it has been among the top-selling liquid CPU coolers since its introduction in late 2012. Note that an upgraded version has recently been released, the Hydro H100i GTX.
Among the many factors that PC builders might consider include the obvious, such as price and performance, but also noise level, size, ease of installation, weight, fan quality and build quality, and we’ll touch on all of them in this article. The table below provides all the vital specifications for our six entrants, along with our installation notes on each one:
Before even looking at performance benchmarks, we can draw a few important distinctions between these coolers from their specs that will be quite relevant to many PC builders. First of all, the Intel cooler is compact, lightweight, and incredibly easy to install, requiring no access to the back of the motherboard, nor any tools. It utilizes plastic push-pins to snap into the motherboard’s mounting holes, and while these can be a bit tricky for first-time builders to fully lock in, once you get the hang of it, the installation of this cooler quite straight-forward. The three Noctua coolers all use Noctua's SecuFirm2 mounting system, and it's by far the best we’ve seen, but the larger the cooler, the more cumbersome it gets to install; while the NH-L9x65 was very easy to install, the NH-D15 was a serious hassle to install. The Thermalright and Corsair coolers were overall the most difficult to install, in part because their mounting brackets just aren’t that well engineered, and the Corsair cooler in particular requires quite a lot of juggling of components and additional cable management to handle its cornucopia of fan cables, USB cables, and hoses. And don’t get us started on Corsair’s CorsairLink software (available for download here) – sure it’s powerful and effective, but wow, is it clunky! It didn’t even function under Windows 8 until late 2014, long after that operating system was “new.”
Another important distinction between the coolers is weight and size, although once you get into the high-performance coolers, it’s a bit hard to compare these specifications directly, since liquid coolers are separated into multiple components. Put simply, big 140mm air coolers are monstrous things, making them a bit unwieldy to install, and in fact, based on our experience, we’d suggest that they be bolted onto the motherboard while outside of the case, with the whole assembly then lowered into the system. It’s just too hard to attach fans and fan power connectors inside a case. They also pose a slight risk of strain to motherboards. We’ve never seen one crack, but we wouldn’t recommend mounting a +2lb. cooler on an ultra-budget, ultra-thin motherboard. That is one key advantage of the Noctua NH-U12S – being a 120mm cooler, it doesn’t throw its weight around during installation. As for the Corsair Hydro H100i, it’s larger overall than the 140mm coolers we tested, and heavier than the Thermalright Macho Rev. B, but once mounted, its bulk is spread out across different sections of the case, making the overall presentation a bit more elegant. In between the Intel cooler and the high-performance air and liquid coolers is the comparatively-tiny Noctua NH-L9x65, which is small, but not entirely a lightweight, given that it's over twice as heavy as the Intel cooler. Still, it won’t post a risk to any motherboard due to its size, nor will it pose any hardship to the installer – it’s by far the easiest heatsink we’ve ever installed besides the stock Intel cooler.
The last specification that we’d draw attention to is fan size and speed. Generally speaking, the larger the fan, the lower the speed required to achieve a certain level of cooling power (technically measured by cubic feet per minute, or CFM). Therefore, coolers equipped with larger fans typically also have lower-RPM fans, although this isn’t always the case. Manufacturers can boost their benchmark ratings a bit by equipping their coolers with ultra-high RPM fans, which result in great performance when noise isn’t considered. Witness, for instance, the relatively high RPM ratings of the 92mm Noctua fan and the 120mm Corsair fans, both quite a bit higher than the tiny fan on the Intel cooler. Also note that the NH-D15 uses specially-modified versions of Noctua's NF-A15 fan, which run at 1500RPM rather than the 1200RPM of the retail product. All of these examples should raise the suspicions of silence-seekers. Therefore, we tested all of our coolers in two different configurations: maximum fan speed along with motherboard pulse width modulation (PWM) or software-controlled “normal” fan speeds.
Because we needed to use a benchmarking platform that could accommodate all of our coolers, from the smallest to the largest, we utilized our standard gaming benchmarking system. The specifications, along with test conditions and applications were as follows:
- Motherboard: ASRock Z97 Extreme4 (thank you to ASRock and Newegg for providing this review sample)
- CPU: Intel Core i7-4770K (running at stock and 4.4GHz overclock)
- Video Card: MSI GeForce GTX 780 Ti Gaming 3GB
- SSD: Crucial MX100 512GB
- RAM: 2x G.Skill TridentX 2x4GB DDR3-2400 Kits (16GB total)
- Case: Corsair 500R
- Power Supply: EVGA Supernova G2 850W
- OS: Windows 8.1
- Stress applications: Battlefield 4 and Intel Burn Test
- Temperature monitoring software: HW Monitor
- Noise level monitoring software: Sound Meter for Android
- Ambient temperature during all tests: between 64.5°F and 65.5°F
Note that we ran our system with all of its included fans plugged in but set to low speed. This included dual 120mm intake fans, a 200mm side intake fan, and a 120mm exhaust fan. The noise floor of this system is about 37dB, which is quiet but not silent, so that will be the minimum sound level you’ll see in our tests, although most of our CPU fans operate above that even at low RPM. Also note our tests were conducted at an ambient temperature between 64.5°F and 65.5°F. Unfortunately, it’s impossible to maintain an entirely constant temperature, if for no other reason than that benchmarking itself outputs heat that increases the temperature in the room.
An unspoken truth in CPU cooler comparisons is that liquid CPU coolers not only add more cooling potential in the form of liquid coolant; they also add more case cooling potential. It’s impossible to separate out the effect of the liquid cooling system from the additional case fans that liquid coolers effectively provide. Note that while many manufacturers recommend installing radiator fans as intakes, drawing cool air in from outside the case, they do this solely to “juice” the benchmarks. As an overall cooling scheme, running radiator fans as intakes is completely illogical. Sure, your CPU will run a bit cooler, but everything else in your system will run much hotter, from the motherboard, to the RAM, to the video card, to the hard drives. Simply put, it’s bad advice, and it exposes the risks of setting up a PC as an amalgamation of parts rather than a unified system. Therefore, we ran our Hydro 100i with its dual 120mm fans set to exhaust, mounted at the top of our case, working in tandem with natural convection to remove heat. But as we mentioned at the start of this paragraph, this added additional case cooling to our system. We considered running top-mounted 120mm fans for our other cooler tests, but ultimately decided that this would simply be unrealistic and unfair – those other coolers did not come with extra 120mm case fans for users to install, and adding such fans wouldn’t be testing the coolers on their own merits. To a certain extent, this illustrates that from a purely technical point of view, liquid CPU coolers are inherently superior, in that they separate the heatsink element from the radiator element, allowing the latter to be mounted much closer to the exterior of the case. That is not to say that they are always preferable, as we’ll demonstrate in our benchmarks. Moving heat from the heatsink to the radiator involves some extra effort, which unfortunately translates to unavoidable noise.
But we're jumping ahead a bit, so let's start at the beginning. First, we need to have a CPU installed before mounting the cooler. You can see below that our Intel CPU is locked in and the CPU area is clear. Note that we’re using four sticks of RAM, a configuration that every cooler should allow for, but not all do, at least not without some difficulty. These are extra-tall, high-performance models from G.Skill, so we’re really using a worst-case scenario here. Indeed, with one of our coolers, installation was sub-optimal due to the height of the sticks, and in another it was all-together impossible without modifying the RAM, as we’ll explain later.
On the next page, we’ll do a deep dive into the installation of each of our six coolers.