A solid performance putting it midway between air and a good custom water cooling loop. It looks good and ease of setup and use are great. The TIM is good and the high speed fans mean it can perform well, however the fans are loud and for high load applications I'd look at going push/pull with AP15's. Shrouds are not worth the cost unless they're free.
The "Sealed" or "All in one" liquid cooler has been growing in popularity over the last few years. Those who can't afford a custom loop or are worried about reliability have often chosen these units. The performance of previous units was usually similar to the high end air coolers such as the noctua dh-14, however the Corsair H100 is the first such unit to feature a 240mm radiator suitable for 2x120mm fans. I had heard rumours that HardOCP had compared it favorably to a custom water cooling setup. However this review is the only one I can find and it doesn't make such a comparison. I myself was interested to see how it does compare to a basic custom water cooling loop and here are the results.
Packaging and aesthetics
Corsair always handles these two well. The box is pretty to look at and the cooler is well protected from damage. All wires are black so no sleeving is required to cover up any clown colored cables. Over all it's tasteful. My only complaint was that the backplate wasn't very specific for socket 1366, it works, but it could have been better. At this price they should have included a block that fits standard metal block that sit on the rear of most cpu sockets.
The CPU block base is polished copper and comes with TIM pre applied. It's not a perfect mirror but it's better than many cooler's I've seen.
The centre of the block has a button that allows you to choose between 3 profiles for the fans. Each setting takes input from a thermal sensor in order to determine the exact fan speed given the load. In my tests I only used the high setting as I'm only interested in maximum performance.
The tubing from the block to the radiator is about 12" long. This is fine for most users, however given that this is a 240mm radiator, some extra length would give more flexibility in locating it given that not every case has a 240mm mounting option within that reach.
|Here is the H100 installed with Gentle Typhoon AP15 fans in push/pull, the tubing is just long enough to make it past a standard size graphics card|
In screwing down the mount there was no spare thread left to screw. In other words I tightened it down until I couldn't tighten any more. With most coolers you tighten until it's tight enough. Luckily it seemed to have exact enough tolerances for this not to be a problem, but it wouldnt' surprise me if someone did have a problem with this.
2 fans were supplied and were recommended to be put in a push configuration. This is not surprising because they run up to 2600rpm and high speeds usually favor push. When at max speed though they are extremely loud as you might expect.
Overall I think Corsair did well with the look, simple, pretty minimal and will match everything.
All tests were completed on an i7-990x running at 4.4GHz (23multx193bclk) with 1.412Vcore and 1.35Vqpi, memory was run at 1600-9-9-9-27-1N. Hyperthreading on, LLC enabled, all throttles, speedstep etc were turned off. Loading was tested using stresscpu v2 for linux 64 bit under redhat 5.7 kernel 2.6.18-274. Stress cpu is based on the gromacs core (from folding at home) and is fairly stable vs time wrt to temps. Temps were logged on every cpu core every minute using kmod-coretemp and lm-sensors lib. The loop was given 10 mins of time to stabilize before logging for 10 mins. The recorded temperature was an average of all 6 core temperatures for 10 minutes i.e. 60 readings. All thermal interface material (TIM) was shin-etsu x23-7723D unless otherwise stated. This was selected as it is very insensitive to poor application and still has decent performance. To be really fair I should have remounted each cpu block 5 times and taken the best result, shin-etsu seemed the best option given that I didn't have time to do this.
It would have been nice to compare this to a loop with an xspc 750 pump from the xspc kit as well as an rs240 radiator and an rx240 radiator to really give a fair comparison.
I tested this as part of a full watercooling test, you can see the full results here.
Fans: A lot of people have asked how the gentle typhoon ap-15's compare to the corsair fans. The short answer is that for maximum cooling the corsair fans are about 2 degrees or 6% better, simply because they run at 2600rpm vs the ap-15's 1850rpm. However they are extremely noisy, too loud in my opinion for a situation where you constantly load the cpu like folding for example.
Push/Pull vs Push: With such a thin radiator I theorized that push would have similar performance to Pull, however again you can see about 3 degrees or 10% better performance with Push/Pull. Note that with push/pull the AP15's perform better than the stock fans in push.
|Test setup - H100 with AP15 push/pull + shrouds|
Shrouds: Adding a 25mm fan shrouds on both sides of the radiator out of old dead fans gave a tiny performance increase of only 0.2 degrees. This was definitely surprising.
TIM: The stock TIM was preapplied and faired a bit better than my slightly smaller than pea sized amount of shin-etsu 7723D, this is probably because I don't get good coverage to the extreme corners of the CPU with the small pea method. I wouldn't bother replacing the TIM unless you're going to use something like Indigo Xtreme. Here's a shot of the stock TIM after removal:
Vs Custom Loop: Here I selected 2 basic loop configurations. The PMP-450S at minimum voltage is roughly equivalent to a D5 vario on setting 4 according to martin's liquid lab. The D5 pump is a very common choice in custom loops however it does dump heat in the loop and so perfomance of a loop with a DDC in it may show better temps. Also a fairer comparison would have the AP15's in push. You can see that custom loops as expected do a good deal better than the H100, however of course they cost more.
|Custom loop testing|
What does this really mean?
Some of you are more interested in maximum overclock than necessarily reducing temps. By making some estimates we can therefore come up with an approximate change in overclock by doing some math.
- CPU power = 200W
- voltage required for overclocking is linear with frequency (which it isn't) and was based on 2 points - 1.4V for 4.4GHz and 1.5V for 4.6GHz
- That the cpu power load is entirely dynamic switching power that follows the standard cmos power equation 1/2 * cap * freq * volts^2
- The max "safe" temperature is 50C above ambient, voltage however can be increased without worry
Using this gave me a formula that from experience looked to be off by a factor of two, applying a correction factor gives us the following plot:
So in other words switching from a bad air cooler to a stock H100 could net you 200MHz. Switching from a stock H100 to a custom loop might net you another 200MHz. This all assumes you are comfortable running the vcore necessary to get there.
The H100 looks good and performs well. The TIM is good, the fans perform well and overall it pretty much hits the middle spot between a good custom loop and an air cooler. For the price that's very good. My only complaint is that the stock fans are so loud. If you're looking to run a load that will require the fans running at max speed a lot then you should really look at changing the fans out.