Battle of the elements: air vs. water CPU cooler showdown

Sponsored by NCIX
Cooling a modern quad-core processor, which has hundreds of millions of gates switching at frequencies north of two or even three gigahertz, is no trivial task. That’s especially true for PC enthusiasts who enjoy pushing their processors past stock frequencies. Instead of keeping just one core in check, all four need to be stable at a given clock speed. As the number of cores increases, so do the odds that one will require extra voltage to keep up with the others, making a good cooling unit even more important.

A typical CPU cooler uses heatpipes to transfer heat from a large, solid metal base to an array of thin, metal fins. For several years now, this approach has yielded better results than one-piece units whose fins rise up from the cooler’s base. Heatpipes have also allowed cooler makers to build massive towers of cooling fins, providing room for larger fans than were feasible with older designs. This brute-force approach relies on plenty of airflow and a large surface area to achieve the best results.

Interestingly, water cooling actually works in much the same way. A water block takes the place of the metal base found on air coolers, circulating coolant though internal channels to pick up as much heat as possible from the processor. Water-filled plastic or rubber tubing takes the place of a metal heatpipe, directing coolant to a radiator that transfers excess heat to the surrounding air. A powered pump pushes coolant through the system, and fans are typically used to generate airflow through the radiator.

Air and water cooling are similar in principle, but the added complication of routing coolant tubing, securing connection points, and mounting a radiator and fan typically makes water-cooling units much harder to install. The cost of individual parts for even a basic water-cooling system is much greater than what you’ll pay for an equivalent air cooler, too. However, CoolIT Systems has introduced an all-in-one Domino ALC water cooler that costs about as much as a high-end air-based unit, and should be much easier to install than a typical water-cooling setup. I wouldn’t be a case and cooling enthusiast if I didn’t jump at the opportunity to test this new Domino. To face it, we’ve rounded up a top-notch air cooler from Noctua, a more traditional and affordable air tower from Kingwin, and a stock Phenom II heatsink. Let’s see how each handles a Phenom II X4 940 at stock and overclocked speeds.

Kingwin’s Revolution RVT-9225 HDT
Heatpipe tower goodness on the cheap

Manufacturer Kingwin
Model Revolution RVT-9225 HDT
Price (Street) $30
Availability Now

I’ve been using the Kingwin Revolution RVT-9225 Heatpipe Direct Touch as the CPU cooler in our enclosure test system for a while now. At $30, it’s cheaper than most tower heatsinks but a bit more expensive than basic units.

The Revolution’s design is typical of cooling units that have been around since the tower heatsink became popular. Three U-shaped copper heatpipes connect to 45 aluminum fins on one side, pass through a base at the bottom of the tower, and then reconnect to the fins again on the opposite side. With this approach, the Revolution is able to maintain a slim profile even with a fan attached.

Although the RVT-9225 is compatible with LGA775 and AM2/3 sockets, it doesn’t come with retention hardware for Intel’s new LGA1366 Core i7 socket. Kingwin doesn’t sell Core i7 retention hardware separately for the RVT-9225, either.

The Revolution’s LGA775 retention bracket does allow you to change the cooler’s orientation in 90° increments, making it easy to direct airflow toward exhaust ports at the rear or top of an enclosure. However, the AM2/3 retention hardware isn’t as flexible. The Socket AM2/3 retention clip orients the cooler’s path of airflow parallel to the socket’s retention tabs, and while it’s possible to flip the cooler 180° to adjust the direction of airflow along that path, you can’t rotate the cooler 90°. With the Asus M3A32 MVP Deluxe motherboard we used for testing, the Kingwin’s airflow can only be directed up or down in a traditional tower enclosure—not toward the front or rear of the case.

Airflow is provided by a 92 mm plastic fan with seven blades. The fan is secured with soft rubber fasteners to keep vibrations from generating extra noise. A four-pin power connector should also help to keep noise levels low by allowing the motherboard to tune fan speeds based on CPU temperatures.


The Heatpipe Direct Touch portion of the Revolution’s name is well-deserved. Kingwin has shaved down the heatpipes to match the contact plane of the rest of the cooler’s base. Soldered joints between the pipes and base should ensure efficient heat transfer, allowing the heatpipes to channel heat directly from the CPU and from the surrounding heatsink base.

Mounting the Revolution was easy thanks to the fact that it clips right into AMD’s standard Socket AM2 retention bracket—no motherboard removal is necessary. The Thermaltake Spedo enclosure pictured above is particularly expansive, so it’s hard to judge the Kingwin cooler’s size in context. At 5.25″ tall, 3.75″ wide, and 3″ thick, the Revolution is only a moderately-sized tower. It should fit inside most mid-to-full tower cases and be compatible with most motherboards. However, if your mobo has larger heatsinks around the socket or if you run taller memory modules, you’ll want to pay attention to the height of the Revolution’s bottom fin, which sits 1.4″ from the base of the motherboard.

Noctua NH-U12P
The tower of doom

Manufacturer Noctua
Model NH U12P
Price (Street) $55
Availability Now

Although the name might not be familiar to everyone, Noctua is an Austrian firm that’s been producing cooling gear aimed particularly at reducing system noise levels for a while now. Noctua means “little owl,” and according to the company, it stands for intelligent and efficient design over brute force. Large tower coolers with massive fans are about as brutish and forceful as air cooling gets, though, so Noctua has some dazzling to do on the intelligent and efficient design front.

Well-known in the quiet computing crowd for being nearly silent, the NH-U12P is one of Noctua’s largest heatsinks. It sells at outlets like NCIX for $55, and comes nicely packaged with a 120 mm fan and plenty of accessories. Included among those accessories is mounting hardware for LGA775 and AM2/3 sockets, but not LGA1366. Noctua does sell a Core i7 compatibility kit for the NH-U12P that runs about $6 online, but they’ll also send you the kit in the mail if you’ve already purchased the cooler and can provide proof of purchase of the heatsink and socket 1366 motherboard.

The mounting hardware packaged with the NH-U12P includes motherboard back plates for both natively supported sockets. Users also have the freedom to orient the cooler’s airflow path however they’d like, regardless of the orientation of the underlying socket.

To help keep things quiet, Noctua provides rubber strips to sit between the fan and heatsink. Two voltage reducers are also included in the box to give users a measure of fan speed control. An instruction manual, screwdriver, and some of Noctua’s own acclaimed NT-H1 thermal compound rounds out the package.

The heatsink itself is crafted from 37 aluminum fins and four U-shaped heatpipes. Each of the fins is cut in a design that Noctua claims allows for smoother airflow between them, reducing turbulence and noise.

The higher cost of the Noctua gets you not only more overall metal than the Revolution, but also nickel plating throughout, which should keep the NH-U12P looking nice for a long time. While not quite a mirror finish, the base of our sample was free of irregularities that might inhibit thermal transfer.

Between the lustrous nickel plating and the attention to detail on each part of the heatsink, the Noctua heatsink oozes good build quality. Time to check out the other half of the package.

Anyone can throw a standard case fan onto a heatsink, but the 120 mm fan Noctua ships with the NH-U12P has several important characteristics the company claims improve performance. For one, the finish on the front side of each blade is textured to break up air, while the back side is smooth to reduce turbulence. Notches in the trailing blade edges purportedly spread noise generated from air movement over a larger pitch range, lessening the fan’s whine.

Installing the NH-U12P took a lot more time than the Kingwin, because I had to remove the motherboard completely to attach the custom back plate. The NH-U12P does weigh more than 1.3 lbs without its fan attached, so there’s a lot of load for the retention bracket to bear. Getting the fan to stay on with the provided steel clips was oddly difficult. The first side would pop off as soon as I tried to slip the second one onto the fan. After some choice words and a little luck, I finally finished the install.

Since the NH-U12P is such a large heatsink, it’s a good idea to double-check dimensions to ensure your motherboard, memory modules, and enclosure will accommodate the cooler. The Noctua’s lowest fin measures 1.6″ from the base of the motherboard, which is a little more room than you get with the Kingwin. However, the NH-U12P is bigger overall, measuring 6.1″ tall, 5″ wide, and 3.9″ thick, including the fan.

CoolIT Systems Domino ALC
H 2 Oh yeah!

Manufacturer CoolIT Systems
Model Domino ALC
Price (Street) $72
Availability Now

CoolIT Systems names ease of replacement over stock equipment and good performance at affordable prices as its primary goals. It’s fitting, then, that the Domino ALC—the three letter acronym standing for Advanced Liquid Cooling—strives to deliver the performance of water cooling in an all-in-one unit that’s easy to install and sells around $70.

This turn-key approach eliminates at least one of the hassles associated with water cooling by bringing all the necessary components together in a pre-built, pre-filled, and factory-sealed unit. Even better, though, is the Domino’s relatively low price, which allows it to compete with high-end air coolers like the Noctua. The only real difference between the Domino and the air towers we’ve assembled is the latter’s use of liquid to transfer heat from the processor slug to an array of cooling fins, which on the Domino, is an actual radiator.

For all the inherent complexity of a water-cooling system, the Domino has surprisingly few parts. The water block isn’t particularly large or heavy, allowing CoolIT to get away with a relatively simple mounting mechanism. Back plates are included in the box for both LGA775 and 1366 sockets, but not for Socket AM2/3 systems. The provided Socket AM2/3 mounting bracket requires that you already have a standard AMD back plate—something most but not all Socket AM2/3 motherboards use.

Leaks are arguably the biggest problem with water cooling, and you’ll want to know about one before your system gets fried. To combat this issue, the Domino is equipped with a flow rate monitor that keeps tabs on water pressure. If a leak develops, or if coolant flow is obstructed within the system, the Domino will sound an alarm. Ideally, this warning should give a user enough time to shut down his system before a leak leads to magic smoke.

The Domino’s pump is hidden under the control unit

In keeping with the Domino’s drop-in aspirations, CoolIT puts a thermal interface pad on the mirror-like business end of the water block. All of the unit’s tubing is connected and sealed at the factory. Even the Domino’s small inline pump and fan come attached to the radiator.

Remember when I mentioned that not all Socket AM2/3 motherboards have a standard AMD back plate? Well, the Asus M3A32 MVP mobo I used for testing doesn’t have one. Fortunately, the Socket AM2/3 retention clip for the Kingwin cooler fits perfectly over the Domino’s water block, securing it tightly in place with no play.

Visible in the picture above, the Domino has a single control button that cycles through its three modes of operation: silent, normal, and performance. The control unit adjusts fan and pump speeds automatically based on the temperature of the coolant and the mode selected. As one might expect, the silent mode allows for warmer coolant temperatures, the performance mode favors higher fan speeds, and the normal mode shoots for a comfortable middle ground between the two. Because pump and fan control is handled internally, the Domino requires a constant 12V power source—don’t plug the unit’s three-pin power plug into a motherboard fan header that varies voltage to manipulate fan speeds.

Once the block is secure, you can position the main unit up to 7″ away thanks to flexible tubing that holds its shape pretty well. The main unit is small enough to fit comfortably under the PSU in traditional enclosures that put the power supply above the motherboard. For those seeking more exact dimensions, the L-shaped control unit is 6.25″ tall, 5.5″ wide, and 5″ deep in the orientation pictured above.

The Domino is designed to have its fan and radiator assembly mounted over a rear 120 mm case exhaust fan. CoolIT holds the system in place with four rubber nubs that are pulled through the fan mounting holes. With enclosures like Antec’s P180, one also has the option of securing the Domino to fan mounts on a case’s top panel. That arrangement would rotate the control unit by 90 degrees, though, making the coolant temperature display harder to read. Of course, if you don’t have a case window, you won’t be able to see the display at all.

Although I was skeptical of the holding strength of the rubber nubs that anchor the Domino in place, I was genuinely impressed at how securely the unit was mounted in our Spedo. Be sure to say that last sentence aloud the next time you’re in a public place for full effect.

I’ve built several water-cooling systems, and each time I’ve had to worry about mounting the block, pump, and radiator; sizing and cutting the tubing; and clamping each connection point securely to avoid leaks. Compared to all that, the Domino ALC’s simple installation is a dramatic improvement. Amazingly, the Domino is even easier to install than some performance air coolers. I have to commend CoolIT for building a solution that even the most timid of PC enthusiasts should be able to install.

The air and water show

As a basis for comparison, we’re going to include the stock cooler that comes with recent Phenom II processors.

AMD’s stock cooler has a couple of heatpipes, but it’s otherwise quite different from the Kingwin and Noctua towers. Most significantly, the AMD cooler’s airflow runs perpendicular rather than parallel to the motherboard. The low-profile, nine-blade fan is also smaller and thinner than the 92 and 120 mm units on the other air coolers. The stock cooler’s fan will have to spin much faster to move the same volume of air as the towers. The good news is that the cooler’s small size will fit into a much broader range of enclosures, including smaller Micro ATX designs.

My heatpipes bring all the heat to the fan

A copper base forms the foundation of AMD’s Phenom cooler. Two pairs of heatpipes are embedded halfway into the copper base before looping upward to transfer heat away from the core. Almost 60 aluminum fins stretch between the base and the fan.

If you’ve been reading our case reviews, you’re already familiar with the test rig we use to see how well enclosures handle the heat put out by a modern PC. In this round-up, we’re using mostly the same gear to push a Phenom II X4 940 on an Asus M3A32 MVP Deluxe motherboard. Based on which allowed for the higher overclock in each scenario, I’ve used either two sticks of DDR2 memory from A-Data or two modules of Corsair CM2X1024 RAM. In testing, I was able to hit higher clock speeds on most of the units with a higher multiplier but lower base clock. The opposite proved true for the Noctua, and since the Corsair RAM seemed to handle the higher base clock better, I used it instead. An XFX GeForce 8800 GTS 512 graphics card and Enermax MODU82+ 625W PSU rounded out the test system, which was wrapped in a Thermaltake Spedo enclosure.

CPU Temperatures

So how do these coolers actually fare when feelin’ the heat? First, we tested the coolers with our Phenom II at its stock 3GHz speed. The Domino was tested in its silent and performance modes.

What you’re looking at here is log of Everest’s CPU temperature probe over a computing session. First, I let the system warm up to a stable idle temperature before starting the logging. I let the system idle for a further five minutes before turning on Prime 95’s maximum heat test for all four cores. The Prime95 load ran for roughly 25 minutes—long enough for load temperatures to stabilize. Finally, I halted the Prime95 load, returning the system to idle, and let Everest log another 35 minutes worth of CPU temperature data.

Let’s break down some temperature averages for each stage of our testing.

Based on the results above, it’s pretty clear that AMD’s stock cooler can’t keep up with the beefier competition. It does manage to bring the CPU temperature down faster during our cooldown period than the Domino set to silent, though. When set to performance mode, the Domino moves into a virtual tie for the lead with the Noctua. The Kingwin does pretty well considering it’s half the price of the Domino and NH-U12P.

Cranking the clocks

Performance at stock speeds is one thing, but what about overclocking? Here’s the highest clock speed we were able to reach with each cooler while maintaining stability through an hour’s worth of Prime95 torture:

Highest stable clock speed reached

Our particular Phenom II X4 940 turned out to be a good overclocker, posting frequencies as high as 3852MHz with the Domino in performance mode. The Noctua and Kingwin towers nearly reached that top speed, managing nearly 100MHz more than the Domino in silent mode. Before even looking at the rest of the results, the value of the aftermarket coolers is already pretty obvious: they’re good for an extra two- to three-hundred megahertz. An equivalent clock speed bump with retail processor models generally costs more than the price of a decent aftermarket heatsink. However, it’s worth noting that even the stock cooler allowed us to hit a higher clock speed than AMD’s flagship Phenom II X4 955 3.2GHz processor.

As we did at stock speeds, we again logged CPU temperatures as the coolers transitioned from idle to a Prime95 load before a cooldown period.

Overclocked CPU temperature over time

The Noctua blows the competition out of the water (somewhat literally), managing CPU temperatures 6-12 degrees cooler under load than the others. The Kingwin and Domino in performance mode look evenly matched, but the water cooler runs much hotter in silent mode.

Overclocked CPU temperature at idle

Even at idle, putting the Domino into silent mode results in much higher CPU temperatures than the others.

Overclocked CPU temperature under load

The Noctua really comes into its own under load, where it yields notably lower CPU temperatures than the best the Domino has to offer. The Revolution slots in a couple of degrees warmer than the Domino’s performance mode, nearly tying the stock AMD cooler. Keep in mind that the stock AMD config is running a few hundred MHz slower than the others, though.

Overclocked CPU temperature during cooldown

During the cooldown stage, the heatsinks return to the same rankings they had at idle.

Good vibrations

A low CPU temperature isn’t much good if a cooler makes enough noise to annoy. We took sound level measurements for each cooler 12″ away from the front, top, and left side of the case to get the best feel for how much noise each setup makes. The Spedo’s side-mounted 220 mm fan is a little loud, so we removed it to start with an acceptable noise floor that most would consider near silent.

CPU Noise at Idle

With the Phenom left at its stock clock speed and the BIOS set to adjust CPU fan speeds based on the processor temperature, the coolers all performed similarly, with the exception of the Domino’s performance mode. This config made a very noticeable whir, with a hint of a mechanical whine. Our sound level meter detected small differences between the other coolers, but even the stock unit was acceptably quiet to my ears.

CPU Noise under load

With a Prime95 load applied to all four cores, noise levels rise slightly for all the coolers but the Noctua and the Domino set to full-bore. I actually noticed the Domino stepping up fan speeds while set to silent—each change in speed resulted in a higher pitch of the characteristic noise. I didn’t perceive the same change for the Kingwin and AMD units, likely because the Domino’s fan is immediately next to the outside of the case, while the other coolers are a little deeper inside the enclosure.

Overclocked CPU Noise at Idle

Cranking up the clock speed makes only the AMD and Kingwin units louder at idle, and then only by a couple of decibels.

Overclocked CPU Noise under load

Our overclocked load test doesn’t change the picture much. The Noctua is again the quietest of the lot, and it’s not any louder here than it was in previous tests, suggesting that the CPU is being kept cool enough not to trigger fan speed changes or that the fan isn’t much louder at higher speeds. The Domino’s fan is rather loud in performance mode, and given its high-pitched whine, I wouldn’t call the noise level comfortable.


A system’s CPU is arguably only as good as the cooling equipment keeping it functioning at full potential. As we’ve seen today with the Phenom II X4 940, a stock unit will do, but aftermarket options can deliver superior cooling and lower noise levels.

If you want a little extra oomph from your CPU, I feel good about recommending the Kingwin Revolution RVT-9225 HDT. It’s not the quietest or the coolest of the bunch, but the Revolution certainly held its own. At a recession-friendly $30—about half what you’ll pay for the Noctua or Domino—the Kingwin cooler offers excellent value. Keep in mind that the cooler’s orientation options are a little limited with Socket AM2/3 systems, though.

Compared to the other air coolers, the $55 Noctua NH-U12P is a brilliant behemoth. Not only did it produce lower noise levels than the rest, it did so without sacrificing overclocking performance. Most importantly, the Noctua kept our overclocked Phenom’s load temperature more than four degrees lower than the best our water cooler could manage.

Noctua packages the NH-U12P with everything you’ll need short of a Core i7 mounting bracket, and while motherboard removal is necessary for installation, you can orient the heatsink however you’d like. It will be hard for any heatsink to compete with the sheer mass of metal that comes in a product like the NH-U12P. When you factor in the cooler’s smart design, excellent craftsmanship, and impressively silent fan, Noctua’s price premium looks well justified. This cooler is perfect for those who want excellent cooling performance without having to give up quiet computing.

Noctua NH-U12P

June 2009

Those willing to spend a little more can get their feet wet with CoolIT’s $72 Domino ALC. Coming from my last water cooling adventure, which involved mounting, connecting, and sealing a stack of individual components, the Domino’s ease of installation was a dream come true. I only wish the cooler’s noise levels and CPU temperatures were, as well. The Domino produced a harsh noise when operating in its performance mode, and when set to silent, its CPU temperatures were notably higher than those of the competition. Setting the Domino on normal offers a decent trade-off between cooling performance and noise levels, but you still get lower CPU temperatures and less noise with the Noctua, which costs less.

I suspect that the Domino’s skinny tubing is to blame for the unit’s relatively lackluster cooling performance compared to custom water-cooling setups. The tubes don’t offer as much flow as the 0.5″ inner-diameter tubing I’m familiar with from my own custom installs. That said, the Domino still managed to (just barely) achieve the highest CPU overclock of the bunch. With connection points sealed at the factory, there’s no need to worry about whether you’ve clamped the tubing well enough to avoid springing a leak, either.



Thanks to NCIX for making this round-up possible. – The freedom to configure your perfect computer system with a huge selection of in-stock computer hardware.

Comments closed
    • moose17145
    • 10 years ago


      • Tamale
      • 10 years ago

      whoa.. totally wrote that description for the wrong fan! thanks for pointing out the obvious mistake! fixed πŸ™‚

    • thermistor
    • 10 years ago

    #64, 65…yes, agreed. I think you nailed it…It’s just not possible to get an air cooler inside a case that will be of comparable surface area to a liquid radiator as part of a water cooling system. Plus the cooling airflow over a radiator could use a fan of practically any size, whereas the case would limit fan size and air flow, with a Tuniq tower or some other cooler being an example of a practical size limit.

    • thermistor
    • 10 years ago

    Having done applications work for under-hood applications involving radiator, intercooler, oil cooler, HVAC, and other cooler for off-highway applications, I am entirely unsure that any water-system can actually do better than an air system. Let me explain:

    Caterpillar on gas engines had an old JWAC system, they used engine coolant to cool incoming charge engine air. This didn’t work so well, so they moved to:

    LACC…Essentially in my mind an analog to water cooling a CPU. The engine used a water loop to cool incoming engine air. This was sorta OK, except that the approach temperatures were not so great, but better than JWAC. You cool the incoming air with liquid, but turn around and cool the liquid with atmospheric air via a radiator. The next, and best step was:

    ATAAC…Air-to-air charge cooling. Cool intake air via a heat exchanger with atmospheric air. It’s the only way their gas engines could get top efficiency and the coldest, densest charge air. Sport tuner guys call charge coolers ‘intercoolers’, but the are the same thing.

    I see a water cooling loop as an intermediate temperature reservoir, and thus less efficient than an air cooler…please point me to something that refutes this thinking.

      • MadManOriginal
      • 10 years ago

      One of the biggest advantages water coolers can have is larger surface area for heat dissipation. It’s something that this particular watercooler and most 1x120mm fan-sized radiators lack, they are at best the same surface area as a large tower-style cooler. You really have to get at least 2x120mm-sized radiators to have an appreciable increase in surface area.

      • Tamale
      • 10 years ago

      ah, but getting cool air into an engine for combustion AND keeping the engine itself cool isn’t really the same thing as simply using air to keep a part cool.

      this is why water cooling can potentially be much better than air cooling in a computer.. the radiators can be much larger.

      However, this could change if you had an air heatsink that somehow attached to the processor and used heatpipes to get the heat over to a much larger radiator… in which time the differences between water cooling and air cooling would basically be gone

      • jackaroon
      • 10 years ago

      I hope so, because I just ordered the new one. 4 pipes instead of 3, same price, looks mostly the same, but the fins start a little closer to the base. Supports the same CPUs, if not more; same size fan (I hope it’s the same fan), still has a PWM (4-pin) hookup . . . lets hope it’s not just to disguise some cost-saving reduction in effectiveness.

      • jackaroon
      • 10 years ago

      The xt-964 is significantly different. It’s good enough for me, but it’s not great. The fan on high is pretty loud and on low it’s probably mid/high 20’s in db. The base could double as a nail file. I don’t really have any frame of reference, for temperature reduction, but it’s a few degrees better than what I had (the intel E6320’s stock unit), which might not be saying a lot.

    • Jambe
    • 10 years ago

    I love the photos! Cooler porn! While I appreciate the novelty of the water cooler, I’d prefer more air coolers in the lineup, even to the exclusion of water coolers. This coming from a fellow who visited the local junkyard for heatercores, mind. It’s just that good air coolers can handle substantial overclocking /[

      • Skrying
      • 10 years ago

      The difference would be small at best. The designers time would be better spent making sure all the heat would go directly to the heatpipes.

    • UberGerbil
    • 10 years ago

    I’m always amazed at how caught up people get with headlines. Somehow a several-hundred-word article is made irrelevant by ten words chosen to be terse and punchy. The intent of a headline is to interest you, to get you to click the link and (start to) read the story, not to give you a comprehensive description of its contents or “state the goal” of the article. Headlines are advertising. Your expectations for them should be in line with that. The quality of the article is in the article, not the headline.

    I mean, I guess they could’ve called this

    “A Comparison Of One Turnkey Watercooler And Two Common (But By No Means Representative Or Best) Aftermarket Air Coolers Without Taking Into Account Much Better (But More Expensive And Time-Consuming) DIY Water Cooling Rigs”

    But I would’ve looked dull and ugly and I would’ve been much less likely to click on it, I think. And people would still be complaining that it was incomplete or misleading, I’d bet.

      • ssidbroadcast
      • 10 years ago

      Actually your headline would’ve been totally lolworthy. I think the problem would’ve been the sheer web-dev rage Cyril would have trying to get that headline to fit.

      And then followed with people complaining that it’s not terse or punchy ; )

      • Meadows
      • 10 years ago


      • flip-mode
      • 10 years ago

      Just call it ‘CPU Cooler Roundup’ and be done. While I agree with you for the most part, calling the review a battle of the elements is slightly heavy handed.

      Titles are important; they are supposed to tell you something; they are supposed to re-present (intentional hyphenation) something intrinsic about the content. Otherwise, books could just be titled ‘Book’ just to be economical, or they could all be titled ‘Chainsaw Massacre’ just to get your attention.

      For what it’s worth, one of my favorite book titles is ‘My Booky Wook’. Think about the poetry of that title.

        • ssidbroadcast
        • 10 years ago

        lengthy title of a Children’s book I grew up with: _[

    • Buub
    • 10 years ago

    Regarding the people who say that this is article is a valid comparison because it’s relatively equivalent cost, you have a point. But not the whole point. Yes, it does meet (somewhat) the goal stated in the subtitle, “How much performance do you really get for your cooling buck?”

    To those (like myself) who say that this article does not meet the goal stated in its title, “Battle of the elements: air vs. water CPU cooler showdown”, we also have a point.

    Yes, cost and ease of installation are very important, and /[

      • Tamale
      • 10 years ago

      All excellent points.. and time / resources permitting, comparing a well-designed custom watercooling setup against these results is toward the top of my priorities for future reviews.

        • Buub
        • 10 years ago

        Awesome glad to hear it!

        And not to make light of the article — it seems well done. But as you see I was looking for something different than what I found. πŸ™‚

          • Tamale
          • 10 years ago

          well, the title and subtext are based on the overwhelming evidence I’ve seen that suggests most people just want to buy something that costs well under a hundred dollars and gets them up and running with their new PC quickly.

          Keep in mind.. if I compared a custom-built kit to the other units in this review and it blew them away, you’d probably see even more complaints about how unfair it was considering the cost of such a system and the difficulties in installing it.. it appears there will always be people who think information is ‘misleading’ simply because everyone thinks differently and expects different things.

            • SomeOtherGeek
            • 10 years ago

            So, for that matter, good job on the review and I, for one, do get the reviews point and it was an awesome read and something that a lot of us will now take into consideration when buying a CPU cooler… The mere thought of easily adding water cooling for cheap is quite mind blowing! Can’t wait to see what will happen in a year or 2 with water-cooling!

            So, once again, good job and keep up the good work!

            • Buub
            • 10 years ago

            Actually, I agree with you. For most people, a modern air-cooled heat-pipe heatsink is the perfect solution. I wouldn’t recommend water cooling for 99% of the people out there.

            My point was simply to make it clear that a good water rig can have substantial performance and/or noise benefits if you’re willing to put in the money and effort.

            But once again, you’re right. I have a water-cooled rig under my desk at home as my main workstation. But all the servers in my garage are running standard heat-pipe heatsinks much like the ones you reviewed in this article. They’re quite good enough, they’re inexpensive, and they’re low maintenance.

            • Lans
            • 10 years ago

            I thought the article was good and the subtitle explained it all and it looks like are suggesting a water cooled rig which that is something I called for too. πŸ™‚

            Still, I do want to make a point, that I wonder what can they do with that additional budget? If anything, can they use silver or something at $100+ or maybe $200+ price levels… Or maybe something more ingenious than just better material for that additional budget? That is my main reluctance in accepting a more expensive water cooled solution as really “better”.

            I still think water cooled rig will be sightly more expensive but will be much closer. With a bit of DIY (copper pipers on hard drives for cooling, etc), you can get very close…. Not so sure about a more extreme DIY approach as I have suggested. πŸ™‚

            Noise might be lower for water cooled rig over all air but there are passive cases near the $150+ range…

            So water cooled vs. passive cases vs. passive cases + (extra 120+ mm fans???) vs. all air rigs… anything else in this battle royale? πŸ™‚

    • Meadows
    • 10 years ago

    I can’t wait for the “earth vs. fire” article.

      • MadManOriginal
      • 10 years ago

      Not that I can be certain but I suspect fire would lose out in a CPU cooling comparison.

        • Meadows
        • 10 years ago

        Show me the graphs.

        • Veerappan
        • 10 years ago

        I disagree…

        1. Fire eats CPU.
        2. Fire runs out of fuel.
        3. CPU remnants return to room temperature.

        Final result: Reviewer sitting on the ground crying about a broken CPU, wondering what ever possessed them to light it on fire.

    • Jypster
    • 10 years ago

    I enjoyed the writeup but I really would liked to have seen a homebrew water cooling setup compared. These are cheap to put together but take some time and thought to get setup. The most expensive part is the waterblock but they can be carried on for years to each machine if you chose right.

      • eitje
      • 10 years ago

      I’d like to see a second article, specifically detailing the setup of a homebrew water cooling system.

      then, run these same benchmarks, and compare with the current article.

    • Usacomp2k3
    • 10 years ago

    Reminds me of the CoolerMaster Aquagate that I got back in the day (and still have in the closet). I had the R120, the 120mm version that performed pretty well. Certainly was much simpler than a full-blown setup, although I don’t know that it performed any better than the stock HSF.
    It was really cool though, and fit the theme of my case with the 4×12″ CCFL’s and 6x80mm LED lights.
    My how things have changed.

    • shank15217
    • 10 years ago

    Nice review, the Noctua heat sink system is the way to go for a decent overclock. I might go for one with the next system upgrade.

    • SnowboardingTobi
    • 10 years ago

    Speaking of coolers… try this:
    Go to Google and search “thermalright”

    Take a look at the first hit… it’s for the thermalright website. But read the umm… “description”…


    • herothezero
    • 10 years ago

    q[< It is no myth. It's fact. Water cooling is superior to Air cooling. Just because you place artificial monetary constraints around the system, does not disprove the facts. <]q There's nothing artificial about the cost--in both time and money--associated with water cooling setups. Pretending not to factor that into the value proposition is absurd. Running a Noctua system on an E8400; silent and stable, even at 4GHz.

      • flip-mode
      • 10 years ago

      At least indicate who you are replying to.

    • Buub
    • 10 years ago

    This is hardly a comparison of water cooling in general. It is a comparison of one low-end liquid cooling solution against several mid to high-end air cooling solutions.

    A Danger Den or Swiftech unit would like give you better cooling under load and/or better noise numbers. But yes, it would be more work.

    • OneArmedScissor
    • 10 years ago

    I am quite a fan of the high end Noctua heatsinks.

    I have a C12P, the “UFO” counterpart of the U12P used in this test, which I use with a Q9400, running 3.2 GHz, with a small voltage boost. With the lowest fan speed, running several passes of Intel Burn Test, the highest temperature the CPU hit was 50C.

    I often run the CPU at near a full load for hours at a time, and it stays in the 40C range.

    I literally can’t hear the fan, at least, not over the PSU fan, which is pretty much the only audible component of this computer. It’s not cheap, but it’s not outrageously expensive, either, and it’s done right in every way, so I have nothing to complain about.

      • Hamish
      • 10 years ago

      I just added a Noctua NH-U9B to my system, and it’s excellent. It’s the little brother of the NH-U12P in the article. I actually stupidly bought that one first without measuring the space I’d need in my case. I ended up selling it cheap to a work buddy who loves his overclocking. He is very happy with it now, too.

    • Sargent Duck
    • 10 years ago

    This comment doesn’t have much to do with the direct article per se, but I’d like to see more of this. Scott/Geoff/Cyril are busy with respected duties and I doubt they have extra time to do this sort of stuff. But Josh, I really think this is where you can come in and really add an extra dimension to TR. Cooling is such an important part of computers but gets so little mention. Go crazy. Do air vs water (like this article), have a big ‘ol cooler shootout, what’s the best thermal compund, what’s the best bang-for-buck cooler, keep tabs on new coolers released, review “other” cooling solutions like fans (S-flex) and what’s the quitest 120mm.

    I really think something like this would be great on TR for somebody to “take over”

    • continuum
    • 10 years ago

    Did you guys list heatpipe diameters anywhere? Just curious, I don’t think I saw it…

    Also this is a pretty diminutive watercooling setup, so honestly, performance was about as expected…

      • Tamale
      • 10 years ago

      no, but I should be able to get those added in pretty easily.. stay tuned.

        • continuum
        • 10 years ago

        Cool, good to know. Since heatpipe diameter is very useful in estimating heatsink performance, seeing it listed is very helpful. =)

    • SecretMaster
    • 10 years ago

    Finally the review comes out! I’ve been waiting months for this since you first mentioned the idea Joshua. I have to dig into this after dinner.

    • flip-mode
    • 10 years ago

    WHOOOOAH! A heatsink review? Holy blue moon Batman!

    Thanks TR!

    • jjj
    • 10 years ago

    The article forgets to metion that Noctua has the NH-U12P SE1366 for the 1366 socket and it comes with 2 fans.
    And the thermal compound is not at all basic it’s a rather good one.

      • Tamale
      • 10 years ago

      thanks.. checking into this.

    • Neutronbeam
    • 10 years ago

    In my experience Noctua baseplates have tiny ridges that do not provide the best surface, but some careful lapping with ultrafine sandpaper gets you a little closer to the copper with a smoother finish. Also, getting the fan on can be tough, but some needlenose pliers work great for manipulating the fan clips.

    • Skrying
    • 10 years ago

    Nice to see a heatsink review on Tech Report. I would like to see two of the more popular options out there right now, the Sunbeam Core Contact and Xigmatek Dark Knight, but this is a good start and Noctua is becoming more popular with their products now available on Newegg (for the longest time they weren’t, and they’ve only been there for less than a year).

    Sucks about the motherboard not giving a proper mounting position. I blame this on a combination of AMD and the heatsink, there are a few heatsink models that get around this issue. Though AMD should really develop a socket layout that allows the user to more easily have the heatsink pointed in any direction… like Intel.

Pin It on Pinterest

Share This