Over the last few years, we've seen an increasing number of liquid coolers positioned as high-end alternatives to traditional heatsink+fan combinations. This has been particularly true in the boutique market, where a number of manufacturers now offer liquid coolers in one form or another.
The image above is from Asetek's website, but all of the liquid coolers we've tested function in the same basic manner. Water is pumped through the CPU cold plate and up to the radiator where a fan blows cool air over the fins and lowers the coolant's temperature. The now-cooled water flows down the return pipe and is again pushed over the coldplate.
|The Contendors: Corsair's H80 & H100|
Corsair currently offers an entire lineup of liquid coolers ranging from the $59 H40 (aluminum heat plate, single fan) to the H80 ($93) and H100 ($115) we're reviewing today. Both the H80 and H100 are based on CoolIT System's ECO II ALC (Advanced Liquid Cooler). Both also incorporate a three-speed fan switch, mounted on top of the pump, that allows users to set a predefined fan speed.
The primary difference between the two coolers is the size and shape of their radiators; the H80 and H100 use the same cooler plate and mounting system.
The H80 has a 120mm radiator that'll mount to any system with an appropriately sized exhaust port in the back of the case. It's designed to be used with up to two fans in a push/pull configuration and pushes between 46-92 CFM depending on fan speed.
Corsair H80 and H100 Copper Cold Plate
The copper cooling plate attaches via four screws and supports LGA1156, LGA1366, and LGA2011 as well as all current AMD sockets. There's a pre-spread layer of thermal paste (which we used). The pump must be oriented in such a way as to ensure that the tubes have proper clearance, which typically means they should point towards either the RAM or the back of the case.
Installation and Setup:
We had no problem installing either cooler, though the LGA2011 mounting screws are a tight fit for the cooler head. Considering how much these kits cost, the included instructions are bare-boned. There's a picture guide without any sort of accompanying text and no user manual, product sheets, or included software. This last, we should note, applies to all of the liquid coolers we've seen recently with the sole exception of AMD's FX LC (not included here for obvious reasons). That kit shipped with fan controller software, while everything else is dependent on physical switches or BIOS controls.
Corsair will sell you a fan control kit...for a cool $100.
|Rogue's Gallery: Competitors from Maingear, Intel|
|Intel's RTS2011LC liquid cooler debuted simultaneously with the company's Sandy Bridge-E and retails for ~$80.
From our review of the Intel Core i7-3986X:
The RTS2011LC features a custom designed 120mmx25mm fan (74 CFM, 21 dBA@ 800 RPM, 35 dBA@ 2200 RPM), a 150mm x 118mm x 37mm radiator, a high efficiency cooler block, and a new heat exchanger design that differed from other Asetek LCLC solutions. According to Intel, at equivalent acoustic levels and at full processor power (130W) the RTS2011LC can achieve a 7°C cooler CPU core temperature than Intel’s highest performance air cooling thermal solution. In addition, the RTS2011LC runs 10 dBA quieter than the high performance air cooling solution when running in performance mode.
Maingear's EPIC-Sized Cooling Solution:
Here at Hot Hardware, we're used to high-end cooling solutions, but Maingear's new liquid cooler is in a class of its own. Maingear calls this behemoth the EPIC 180, presumably because Coolersaurus Rex was already taken. At 7 inches wide, 9 inches tall, and 2.5 inches thick, this is the Hulk Hogan of CPU radiators.
It's hard to get a sense of how big this thing is,
partly because we can't find a way to squeeze a pony into the frame
Mounted in Maingear's SHIFT case, you can see the real estate consumption but the motherboard area is tidy
Despite its size, the Epic 180 uses the same Eco 2 pump as the Corsair models, making this comparison an interesting example of how increasing total radiator size impacts cooling capability. According to Maingear's self-published data, using a much larger fan can have a marked difference on cooler performance even when the same pump is used.
Chart and Data Source: Maingear Computers
Unlike the other coolers listed here, the Epic 180 isn't available for separate purchase. Considering even our Corsair 800D quailed at the thought of trying to contain it, that's likely just as well. We could only test the Epic 180 by rigging a spot beside the case for the radiator to sit on.
|Test Setup, Stock Cooling|
|All tests were conducted using Intel's Core i7-3960X hexa-core Sandy Bridge-E CPU and DX79SI Siler motherboard. Windows 7 was set to 'Balanced' power mode. Idle temperatures were measured after allowing the system to sit post-boot for 10 minutes. Load temperatures were measured after running Intel's stress test as included in the company's Extreme Tuning Utility (XTU) for one hour. Peak temperatures were taken while running Prime95 for two hours, with that utility configured for maximum heat and power consumption.
We used a Corsair 800D chassis to test mounting and fittings, but left the chassis open.
Intel's stress test appears to load the CPU nearly as well as Prime95, but since it's not widely available, we went with Prime as a generally known and easily replicable test.
Our stock speed and 4.13GHz comparisons were done with the Corsair coolers in 'Quiet' mode (lowest setting). Our 4.5Ghz test required us to increase fan speeds; so we brought both coolers up to the 'Performance' preset (highest setting). Note that Intel's cooler uses a four-pin fan and default BIOS settings that focus more on quiet operation than maximum performance. During our stock test, the RTS2011LC's fan never spun faster than ~875 RPM.
So how do things look out of the gate?
Pretty darn good. The RTS2011LC's higher temperatures are more the result of ultra-low fan speeds than any particular problem with the cooler. The H80 outperforms Intel's cooler by a fair margin, the H80 is beaten itself by the H100, which goes down in turn to Maingear's Epic 180.
|Overclocked Performance, Noise|
|In our first overclocking test, we push the CPU's maximum frequency up to 4.13GHz from a 3.6GHz turbo mode (a gain of 14.7 percent). We also increased the core voltage to 1.35v. The H80 and H100 were set for 'Quiet' mode.
Here, the LC2011RTS and Corsair H80 are essentially tied, as are the Epic 180 and H100. Intel's cooler spins up sharply at this point; fan RPM rises to ~2200 up from 800.
For our second overclocking test, we increased the CPU to 4.5Ghz, a further gain of 8.9 percent (25 percent faster than the original Turbo Mode). This modest gain required an additional 0.02v (1.37v CPU). Our CPU apparently tops out right about here -- we weren't able to stabilize the chip at 4.67GHz, no matter what combination of voltages and multipliers we tested.
The Corsair fans were set to 'Performance' for this test.
Intel's cooler wasn't able to stabilize the chip in Prime 95 during our peak test; the program crashed within minutes. The H80 and H100 fared better; the H80's temperatures fell slightly thanks to the increased fan speed while the H100's thermals only increased modestly.
The Epic 180's temperatures puzzled us until we talked to Maingear and realized that the cooler's fan was spinning at the same speed no matter what the CPU load was. When we turned off BIOS control and set the fan to full speed, temperatures dropped significantly. Here, the 180mm design pays off; 1200 RPM fans are generally quiet regardless; a 180mm fan at 1200 RPM is nearly silent.
The RTS2011RC runs into trouble for an obvious reason -- as we push the CPU's voltage upwards, total power consumption rises sharply.
One of the supposed advantages of water coolers is that they're quieter than standard heatsink+fans. Our normally trusty decibel meter isn't capable of picking up sounds below 50 dB, which made it impossible to objectively measure noise levels between the various coolers, but we have a few subjective observations to share.
First, the much-touted aural advantage of liquid coolers is undercut by at least a little pump noise. Even if LC's are quieter on an absolute scale, replacing a modern heatsink+fan with a liquid cooler means substituting the quiet sound of low-speed airflow for the quiet sound of a small motor. Which of these suits you better is a matter of personal preference.
In every cooler we tested, the pump noise was actually louder than the fans when the CPU was idling. Maingear's Coolersaurus tied Intel's RTS2011LC for quietest subjective cooler when the system was idling and took home the bacon as far as quietest overall design thanks to the Radiator That Ate Manhattan. The H100 and H80 followed it, as one would expect. The H100 is quieter than the H80, even at the same fan speed, thanks to the larger radiator that sits two fans beside each other rather than doubling them up one behind the other.
|Performance Summary: Both the Corsair H80 and H100 offer significantly improved performance over Intel's liquid cooler and allow for stable overclocks at a higher frequency than the Intel product can reach.
Between the H80 and the H100, we like the H100 more--provided you've got a compatible chassis and feel like dropping three figures on a CPU cooling solution. Maingear's Epic 180 is also an interesting choice that performs very well, better than Corsair's units actually at some settings, though its not commercially available unless you get a full system from Maingear.
Should you consider dropping $100 or so on a CPU cooler? It seriously depends on what you want. If all you care about is making sure Turbo Mode kicks in regularly and you aren't concerned with overclocking, there are much cheaper solutions (including LGA2011-compatible coolers). Serious overclockers, on the other hand, will find a lot to like in these designs.
Do the Corsair H80/H100 offer performance at a decibel level that air coolers can't match? A bit of independent research online suggests not. Compare reviews around the web and you'll find that there are several high-end air coolers capable of competing with (if not surpassing) the H80 and H100 on both fronts.
The advantage of the liquid coolers, in such cases, is twofold. First, they're often easier to install. Attempting to bolt a huge heatsink to a motherboard that's already inside a case can be a torturous affair, leading to skinned knuckles, harsh words, and failed marriages.
Second, liquid coolers move the heaviest part of the cooler away from all the other sensitive case hardware and bolt it securely to the back, top or bottom of a case. True, it's rare for a properly secured heatsink to be jarred loose and go caroming around the inside of one's chassis. Rare, however, isn't completely unheard of, though the thermal advantages are more interesting perhaps. The real benefit is that the radiator is generally mounted away from other components, so heat is dissipated outside the case, rather than inside the motherboard area, where traditional air coolers can sometimes warm-up other components.
If you're looking for a liquid cooler, Corsair's are both good options--and we'd recommend them over Intel's RTS2011LC, hands down. One thing we'd really like to see from Corsair, however, is a commitment to honor future products with mount kit upgrades, similar to what Noctua announced earlier this year.
At $90-$120, these aren't just single system products -- they're significant investments. We know they'll be compatible through the Ivy Bridge product series, but a prominent guarantee to provide upgrade kits for future sockets would give Corsair an advantage compared to the other liquid cooler vendors.
Corsair's H80 and H100 Liquid Coolers