Cooler Master’s Hyper D92 CPU cooler reviewed

CPU coolers are a relatively slow-moving backwater in the relentless torrent of PC hardware. As Cyril noted in his last review of a CPU tower cooler, air cooling is pretty much a solved problem: bend some heat pipes, solder some fins on, add some fans, and you’ve got yourself a tower-style cooler.

That doesn’t mean that the CPU air cooling market is entirely stagnant, though. Modern PCs are getting smaller, and smaller cases demand smaller heatsinks. While I tend to favor closed-loop liquid coolers in compact PCs, not everyone has the budget or desire to use one, and liquid coolers can have problems that air coolers are free from by design.

Enter Cooler Master’s Hyper D92. This $45 tower is only 5.6″ tall, which is perfect for smaller cases like Corsair’s Graphite Series 380T. The D92 also has 1.8″ of clearance between its base and the bottom of its fin stack, which leaves some room for RAM with tall heatsinks.

To keep air moving across its fins and heat pipes, the D92 uses a pair of 92-mm PWM fans in a push-pull configuration. Here, we can see the most notable feature of the D92’s design: the offset fan setup, which is supposed to prevent stagnant air from gathering between the fan hubs. Beside each fan is a spur of fins that adds to the heatsink’s surface area. Each fan is backed with four rubber pads to keep undesirable vibrations from being transmitted into the heatsink tower.

The fans are secured with a pair of tool-free plastic clips. These clips grab onto notches on either side of the fin stack. Since each fan can only clip onto the tower in one orientation, this system is pretty foolproof and easy to use. Kudos to Cooler Master here.

The D92’s base consists of four U-shaped copper heat pipes that make direct contact with the CPU heat spreader, interspersed with rows of aluminum from the clamping plate above. The finish quality of the base is similar to that of Cooler Master’s closed-loop liquid coolers, with a rough pattern of tooling marks visible to the naked eye.

The “Continuous Direct Contact” base design of the Hyper 212 Evo. Source: Cooler Master

I do wonder why Cooler Master didn’t use its fancy “Continuous Direct Contact” base design on the D92, as it does on the Hyper 212 Evo. The company claims the CDC base is good for a 5% increase in cooling performance versus its non-CDC counterparts, which could be a helpful boost for the smaller D92. Perhaps it’s just the vagaries of product segmentation at work.

Cooler Master includes two universal mounting plates with the D92: one for AMD CPUs and another for Intel chips. The AMD plate supports everything from Socket AM2 onward, while the Intel plate will mate with LGA775 and newer sockets. There are matching screws and brackets in the box, too, as well as a tube of thermal paste and a PWM splitter.

Aside from its offset fan design, the Hyper D92 sticks to the well-worn path that virtually every modern tower-style air cooler treads. The air-cooling market is quite mature at this point, so that’s not terribly surprising. It’s nice to see that Cooler Master is considering the needs of folks with more compact systems, though.

Now that we’ve looked at the Hyper D92, let’s see how easy it is to mount.



One of the most common complaints I’ve seen about the Hyper 212 Evo is that it’s a pain to install. The existence of several YouTube installation guides, some with well over 100,000 views, seems to confirm this difficulty. Cooler Master appears to have addressed some of these complaints with the design of the Hyper D92’s mounting system.

The foundation of the Hyper D92 (on Intel boards, at least) starts with an x-shaped plastic brace with captive screws at its tips. These screws click into one of three positions depending on the socket type. The brace is secured to the mobo with four nuts that sit between the cooler and the board itself. Cooler Master includes a sheet of insulating stickers that need to be applied to the motherboard-facing end of each nut before they’re screwed down. You can see the stickers on the nuts in the photo above.

Once the nuts are in place, a metal crossbar goes on top of each pair, followed by another set of screws. The heatsink tower attaches to these struts with a pair of spring-loaded captive screws that thread into each one. In order to get a screwdriver to these screws, the fans need to be unclipped from the heatsink tower. Once the D92’s tower is screwed down, clip the fans back in place, and you’re all done. I had no difficulty with any part of the process.

If you’re one of the lucky few with an LGA 2011 CPU, a backplate and threaded screw holes are already part of the socket. Cooler Master includes four bolts that screw into the threaded holes, atop which one mounts the same struts used with the other sockets. From there, installation is identical: just screw down the heatsink and clip the fans back into place. I don’t have an LGA 2011 motherboard handy to test this process, but it looks even easier than for LGA 1150.

Now, let’s see how the Hyper D92 stacks up against Intel’s stock cooler—and how it handles an overclocked CPU.

Our testing methods

To test the Hyper D92, I’ve strapped it onto a Core i5-4690K. This CPU’s unlocked upper multiplier allows for easy overclocking, and I’ll be taking advantage of this freedom to make the Hyper D92 sweat.

I’ll also be running our test sequence with the Core i5-4690K’s stock Intel cooler. This cooler should provide a useful frame of reference, particularly on the noise front. TR’s most recent hardware survey shows many enthusiasts use aftermarket cooling even if they don’t overclock, which tells me that noise reduction is a key concern.

Here’s the full configuration of the test system:

Processor Intel Core i5-4690K
Motherboard Asus Z97-A
Memory 16GB Crucial Ballistix Sport DDR3-1600 (2x 8GB DIMMs)
Graphics card EVGA Nvidia GeForce GTX 760
Storage Samsung 840 Pro 256GB SSD
Power supply Seasonic Platinum Series SS-660 XP2
Wireless networking Intel Centrino Advanced-N 6205
OS Windows 8.1 Pro

Our CPU cooler testing regimen is as follows:

  • 10 minutes idling at the Windows 8.1 desktop
  • 20 minutes of the Prime95 Small FFTs CPU torture test
  • 10 minutes idling at the Windows 8.1 desktop

Data was logged using AIDA64 Engineer. To rule out case cooling as a factor, tests were run on an open test bench. The ambient temperature at the time of testing was about 74°F (23.3°C).

I plugged the D92’s fans into the CPU header on the Asus Z97-A using the included PWM fan splitter. Fan speeds were controlled by the motherboard firmware’s “standard” profile for both the D92 and the Intel cooler.

The tests and methods we employ are publicly available and reproducible. If you have questions about our methods, join us on our forums to discuss them.


Stock-clocked cooling performance

Here are the test results from each cooler, plotted over time:

And here are some minimum and maximum numbers from each testing phase:

The Hyper D92 performs much better than the Intel stock heatsink under load, delivering an impressive 21°C drop in CPU temperatures. The 79°C peak with the stock cooler is fairly toasty, but it’s still below the CPU’s throttling threshold. Keep in mind that our Prime95 load produces more heat than typical desktop workloads.

We shouldn’t be surprised that the Hyper D92 is a superior cooler to the stock unit: it has an extra fan and much more surface area on its side.

Noise levels

I used the iOS app dB meter to measure the amount of noise produced by each heatsink. My iPhone isn’t a lab-grade instrument, so take these results with the appropriate grain of salt. In order to isolate the noise of each cooler, I removed the system’s graphics card and switched to the Core i5-4690K’s integrated GPU for this test.

Noise levels were measured with the iPhone’s microphone positioned 6″ from the top of each cooler.

As the results show, the D92 is substantially quieter than the Intel cooler under load. While the stock unit is barely discernible at idle, the character of its sound at full speed is coarse and unpleasant, with a distinct rattly hum.

The Hyper D92 doesn’t deliver a decisive victory, though. At idle, its twin 92-mm fans produce a faint hum with a somewhat growly character. The sound seems to be coming from the fan motors—it’s not simply a consequence of having an extra fan. I’d actually say the tone produced by the Intel cooler is preferable. Where the D92 does stand out is under load. Even with Prime95 running at full tilt, the D92 only gets one decibel louder, and its noise character doesn’t change appreciably. I’d wager that putting the D92 inside a case and under a desk would make the hum much less noticeable.

Overclocking performance

To test the D92’s limits, I turned the screws on the CPU. After the usual trial and error, I arrived at a stable frequency of 4.4GHz, with the CPU pulling an indicated 1.216V under our Prime95 load. (Asus’ Windows utility reported a slightly higher voltage than the 1.190V I set in the firmware, but we’ve seen similar discrepancies on other motherboards.)

The system was stable at these settings, and the CPU hit a peak temperature of 80°C. At higher speeds, instant blue-screen errors appeared unless I added more voltage—but since higher voltages caused CPU temperatures to rocket toward 90°C, I held off on pushing any further.

4.4 GHz isn’t a bad result for a quad-core Haswell CPU combined with a compact cooler like the D92. Of course, given the inherent variance between individual chips, your mileage may vary.

The Hyper D92’s twin fans spun at nearly their full speed with the overclocked CPU under load. I measured noise levels at a distracting 55 dBA, and subjectively, the fans sounded something like a hair dryer in full song.

That said, I squeezed a decent amount of extra frequency out of my CPU. If you’re after a more modest overclock, have a chip capable of running on less voltage, or want to overclock a less demanding CPU like the Pentium Anniversary Edition, the Hyper D92 may not have to work as hard to keep things cool.



We’re often told that there’s no replacement for displacement, whether in internal combustion engines or CPU heatsinks. Does the Hyper D92 deliver the goods despite its diminutive stature?

On a stock-clocked CPU, the Hyper D92 is a vast improvement over Intel’s included cooler in nearly every respect. The D92 is whisper-quiet both at idle and under load, maxing out at only 30 dBA in my tests. The D92 also lowered load temperatures by an impressive 21°C compared to the stock heatsink. I found the mounting system easy to use, and I appreciated the foolproof, tool-free fan brackets, too.

My one complaint lies with the D92’s twin 92-mm fans. I wish these spinners were more refined. On a stock-clocked CPU, they don’t have to spin up much to keep things chilly, so the only flaw is a minor growly character that will likely be dampened by a case. The D92 gets quite loud when its fans are at full speed, however, turning in a 55 dBA result in my overclocking tests. That’s probably too much noise for most people to tolerate.

All told, if you need a relatively compact heatsink for your next microATX or Mini-ITX build, the Hyper D92 is a solid upgrade over Intel’s stock cooler. If you’re willing to put up with the possibility of higher CPU temperatures and more noise than you might get from a bigger tower, you can even overclock fairly well with the D92. I was so impressed with this compact tower that I purchased one for another build of my own, so I can confidently say that the Hyper D92 is TR Recommended.

Did you enjoy this review? Pay what you want to subscribe and support us.

Comments closed
    • VincentHanna
    • 6 years ago

    CPU cooling is a “solved problem”
    Replacing the stock cooler with this(or anything else) will dramatically improve your numbers.
    Here is one example of (or anything else), see, it looks shiny.

    • Chrispy_
    • 6 years ago

    You guys need to ship Jeffrey some hardware to test alongside.

    The real question is not “is this better than a stock cooler?” but “is this better than everything else on the market?”

    Otherwise, this puppy looks good. I’m moving away from humongous tower coolers because chips run at 4GHz+ on stock voltage with stock cooling, there’s just no need to hang 2lbs of metal off your processor socket any more.

    • tipoo
    • 6 years ago

    The base of the unit, just visually from the pictures,looks the same as the linked one with the continuous direct contact. What is the difference supposed to be? Just the amount of smoothing on the bottom? In terms of placement of the copper heatpipes through an Al block, they look the same.

      • JustAnEngineer
      • 6 years ago

      On the [url=<]Hyper 212 Evo[/url<], there are no gaps between the heatpipes. They're squished together and flattened to make the contact area for the processor. Click on the third photo in the gallery. On the [url=<]Hyper D92[/url<], there are gaps between the heatpipes where the aluminum block sticks through to make contact with the processor. Click on the fourth photo in the gallery.

        • tipoo
        • 6 years ago

        Pff, the picture I was looking at in the article was of the first one with CDC. Doh.

    • _ppi
    • 6 years ago

    I am sorry, but I am highly disappointed with this review. Which I find highly surprising given the otherwise high quality of content on this site.

    Comparing aftermarket cooler ONLY with Intel’s stock one? Where’s the real competition? Reading noise with iPhone?!?!?!?! C’mon guys, you can do better.

      • NIKOLAS
      • 6 years ago

      Yep, this is bordering on literally being a useless review.

      This is also not the first time in recent months I have seen a pointless review like this.

      Last time I didn’t say anything, but this time it must be said.

      • Jeff Kampman
      • 6 years ago

      I’m sorry that you didn’t enjoy my review. A couple of things:

      Including the Intel stock cooler as a point of comparison is not meant to be any form of competition, it’s there merely to provide a frame of reference for my results. You’ll note that I didn’t consider the results surprising; anybody who’s bothered to replace a stock heatsink with an aftermarket one knows as much. Intel’s stock coolers are also ubiquitous: while not everybody has experienced a cooler from Noctua, Cooler Master, Cryorig, Phanteks, or the like, pretty much everybody with an Intel PC has heard the stock heatsink in operation.

      Secondly, I’ve used the same noise measurement methods in every review I’ve written. We have been entirely transparent about this practice. If the editors at TR believed that my methods were unreliable or misleading, they would not allow me to publish my data.

      In the spirit of inquiry, however, I surveyed some of the other D92 reviews available online. In the one (!) other review where noise levels were actually tested, I found that my results were within 2 dBA of the other site’s load numbers, while the idle noise they recorded was just above the noise floor of their meter (35 dBA.) You can take what you will from that, since our methods and testing environments aren’t directly comparable, but it doesn’t make me worry that my numbers are wildly inaccurate—if anything, it suggests that the iPhone is a pretty good noise meter in a pinch.

      We’re working on expanding our coverage of CPU coolers, so I hope to have a bigger stable of hardware to compare to with time. I hope you’ll continue reading despite my greenhorn limitations. If you’d like to further this progress, please consider subscribing.

        • rechicero
        • 6 years ago

        If you want to expand the coverage of CPU cooler, might be useful not using open rigs. Using an open rig for testing a cooler is like testing high end sound equipment with a low bitrate mp3 in an airport. Everything seems the same because of the conditions of the test (best case scenario for bad coolers).

        Use the simplest of cases, with the standard 1 frontal low fan-in, 1 posterior high fan out, and with the 4 possible scenarios. That would be useful. And better, with GPU with custom cooling and let’s see how the coolers actually cool.

        I bet you’ll find more differences. Specially on the sound front.

        Open rigs are absurd. And I know you can racionalize it with “everycase is different” claims… And that’s true. But a simple case will be always more similar to any other case than an open rig.

        • VincentHanna
        • 6 years ago

        I think you missed the point.

        When you can remove the title of, and images of one CPU cooling device, and replace it with literally [b<]any other CPU cooler[/b<] sold in the past 10 years, that is a problem. [b<][u<]At the bare minimum[/b<][/u<] include [list<] 1 or 2 more air coolers from the same price range/school of design 1 popular/ widely selling CPU cooler that people are probably familiar with 1 more expensive/premium cooler 1 Closed loop cooler [/list<] You might consider an oblique reference to TRs 2014 system guide, a "site favorite" cooler (etc).You should also talk about the quality/specs of the stock fans (independent of the main testing), some coolers basically require a fan upgrade upon purchase. Leave the stock cooler in the review, its a good baseline. If not for the sake of having a content rich review, then do it for the sake of the google web crawlers who are scouring your site for keywords and just looking for a reason to GIVE YOU FREE MONEY!

    • vargis14
    • 6 years ago

    Is there any way Jeffery or anyone at TR can do a comparison test between probably the most used AM cooler the 35$ Cooler Master Hyper 212 EVO against Cooler Masters 212’s Bigger brother the $46 Hyper 612 Ver.2 with its 6 continuous direct touch heatpipes .

    The 612 is even capable of cooling a 95 watt CPU passively and they designed it with bends in the heatpipes so the fan is not in the way of tall memory heat spreaders like the 212. Also from those bends the fin stack is more over top of the IO portion of the motherboard and close to most rear exhaust case fans.

    I have read a couple reviews they used the Hyper 612 to cool a Intel i7-5960X @ 5.0GHz and it kept the CPU at 76C I think that is pretty impressive for a 46$ cooler and a single quiet 120x120x25mm model which has RPM of 800-1300 and it uses a 4pin PWM connector. The air flow rating is 26.6-44.2CFM (45.2-75m3/h) with air pressure of 0.6-1.6mm H2O and it uses a rifle bearing design with curved fan blades. Our noise level rating is 11-20dBA

    • slaimus
    • 6 years ago

    Can you post motherboard/VRM temperature comparison between the two heatsinks? I have been told to stick with the Intel stock cooler because the tower-type coolers do not blow any air towards motherboard components like the VRM, whereas the stock headsink is design to direct some of the air off of the heatsink to cool the motherboard as well.

      • continuum
      • 6 years ago

      Also, could we get some fan RPM data? I noticed it was missing for the OC’ed speed– although you guys did report noise, which was appreciated. Thanks!

        • Jeff Kampman
        • 6 years ago

        I have the system back in a case with more fans now, but I loaded the same OC profile and Prime95 load again to see what I could find out. The AI Suite 3 utility reports that the CPU fans spin at about 2700 RPM under those conditions, and it sounded the same as it did on the test bench. I’ll be sure to log more in-depth data in the future.

      • Jeff Kampman
      • 6 years ago

      For what it’s worth, the motherboard temperatures I recorded were within 1°C of each other during each testing phase: 24/27/24°C for the Hyper D92, and 24/27/25°C for the Intel heatsink (idle/load/cooldown.) I didn’t have any additional fans blowing on the motherboard, and I didn’t notice any system instability during my stock-clocked testing or while overclocking.

      I think that so long as you have an exhaust fan of some kind near your CPU socket, you’ll draw enough air over the VRM heatsinks to prevent any instability. I didn’t have any extra fans blowing on the motherboard during testing.

      • Firestarter
      • 6 years ago

      that would only be an issue if you’re overclocking, which you won’t be doing with the stock heatsink. Besides, some high end Intel CPUs used to ship with a tower cooler too: [url<][/url<] I wouldn't worry too much about it

    • Firestarter
    • 6 years ago

    honestly, when looking at something to replace the stock cooler with 0 thoughts about overclocking, I’d just take the cheapest tower cooler that fits. At $45, this one definitely does not qualify

      • derFunkenstein
      • 6 years ago

      Yup. 212 EVO still wins for me. Hyper TX3 if you don’t mind smaller fans.

      • LoneWolf15
      • 6 years ago

      Agreed. Got the Arctic Cooling Freezer 7 Pro for the wife’s system when it was on sale. Less than half the price, fits in a micro-ATX case, and keeps an i5-2500 plenty cool.

      EDIT: The Freezer i11 is a better deal now.

      • My Johnson
      • 6 years ago

      The 212 EVO is pretty darn cheap. I recall paying $22 for it, and a search shows prices have risen just a little bit:


    • JustAnEngineer
    • 6 years ago

    I’d really like to see a few comparisons to the similarly-compact [url=<]Hyper TX3[/url<], which ships with just one 92 mm fan and has just three heatpipes compared to the D92's four. If you were clever enough to purchase a case that can handle the taller heatsink, the 120 mm fan on the Hyper 212 Evo is going to be less obnoxiously noisy than any pair of 92 mm fans.

      • albundy
      • 6 years ago

      yup. also any decent cooler will be better than stock. it always has been…always will.

      • DPete27
      • 6 years ago

      I too was thinking it would have been nice to see a comparison to the TX3 and probably also the [url=<]N520[/url<] which the D92 [b<]DIRECTLY[/b<] replaces.

      • flip-mode
      • 6 years ago

      This. Hyper TX3 is $20, less than half the cost of the D92.

        • JustAnEngineer
        • 6 years ago

        The inexpensive 92 mm Cooler Master [url=<]Hyper TX3[/url<] is [url=<]$18.75[/url<] at Amazon or [url=<]$22[/url<] at Newegg. The similar Arctic Cooling [url=<]Freezer i11[/url<] is [url=<]$23[/url<] at Amazon or [url=<]$24[/url<] at Newegg. Neither one of these is going to provide as much cooling for extreme overclocking as the [url=<]Hyper D92[/url<] in this review, but are they "good enough"?

Pin It on Pinterest

Share This