This will be a 2 part round-up of the following coolers:
- CoolerMaster Hyper 212+
- Corsair H50
- Noctua NH-D14
and following pastes:
Using the results here, along with the many benches and reviews out on the web, I think you can draw conclusions about the products discussed here. This is not meant to be an exact study on the products, but rather a relative one, that will answer the following questions:
Warning: Spoiler! (Click to show)
- What is the difference between a low end, a mid-range, and high end, CPU cooler?
- What is the difference between an established, high-end paste like PK-1, and the new generation of pastes like PK-3?
- How good is PK-2 and PK-3 (given that we pretty much know exactly how PK-1 compares to every other paste out there)?
- Can buying high end paste result in a significant drop in temps for a very good price to performance ratio, or is it just an unnoticeable degree or two at best?
Warning: Spoiler! (Click to show)
I attempted to strictly stick to apples to apples comparisons here. As I see it, testing heatsinks at stock is really quite useless - a certain heatsink might be awesome, but due to a low quality fan (perhaps, so as to make sure the price is quite affordable), the heatsink is written off as low quality. Or, perhaps the heatsink comes with a fan that is focused on being quiet, or long lasting, instead of being powerful.
For example, the Corsair H70 comes with 2 powerful fans (that still suck), and the H50 comes with 1 weak fan. The H50 actually outperforms the H70 in apples to apples, it has more FPI and is a better cooler, but with stock set-ups, the H70 wins simply because it has 2 fans (with closed loops, whoever has the biggest fans will do the best anyways, moreso than design). Both H50 and H70 come with crap fans, and you need to replace them to get their real performance anyways.
Or, take the Hyper 212 Evo. It tends to perform about 2-3*C better than the Hyper 212+. Now, both of these fans come with absolutely atrocious fans, that are deafening, making an annoying whine, and extremely powerful. This is partly why these heatsinks are so cheap and cool so well. However, the Hyper 212 Evo, has a slightly stronger fan, slightly more FPI, and a lapped base. In apples to apples tests, the Evo is only maybe 1-2*C better, which is basically the results you'd get for lapping the base - in short, the Evo is just a 212+ with a lapped base and a better included fan, which is not worth the ~$9 price premium (getting better thermal paste with that $9 would give you a bigger temp drop).
And we're all enthusiasts here. Chances are, most of us have a huge stack of fans anyways, or we throw away the stock fans because even on a mid-range or higher end cooler, they still are crap, overpriced, and loud. So this review is focused on true heatsink performance, not on how good are the fans included with the heatsink.
Because the real point of this review is not Hyper 212+ vs H50, but Low End vs Mid Range vs High End cooling, and what fan a heatsink comes with varies. I want to see, is a Megahalem/H60/Mugen/9900MAX a significant temp drop from the Hyper 212+, or is it not really any different? Is a dual tower worth the huge price tag, or is it just a tiny bit of performance for a ton of extra cash?
So with a focus on apples to apples, here's how I tested the heatsinks:
- Phenom II 955 Overclocked to [email protected] Everything else stock.
- PK-2, as it is very easy to apply (marketed as easier than PK-1 and PK-3, and I agree), with a rice grain amount.
- Tests are done in an air controlled room, with consistent case temps
- On the Hyper 212+, 2 x Yate Loons Mediums @ 100%. Spreading paste was done with spread method.
- On the H50, 2 x Yate Loons Mediums @ 100%, as intake, with the radiator in the front of the case. Normally, I cable tie an intake fan in my 5.25 in. drive bays, but I took the H50 and cable tied it there in place.
- On the NH-D14, stock Noctua Fans 120mm + 140mm. The best for apples to apples would be Yate-M/140mm/Yate-M, to include the advantage of 3 fans and being able to use a 140mm pressure, but the NH-D14 was so much better even at stock configuration, Yate Loons are unable to use the Noctua's wire mounts (without heavy modding).
- Case temp 25*C, 69.5-70*F ambient
- Prime95 Custom Small FFT Min length 8/Max Length 8, Priority 10/Above Normal, for at least 20 minutes. IBT is inconsistent and gives are sort of plateau, then rise in temp, for each new test run, and Blend does not stress the CPU like Small FFT. Min/Max FFT length of 8 maxes out the CPU temps and keeps results consistent.
- NZXT Gamma case, 2 x Intake (bottom, top cable tied in), 1 x Rear, 2 x Top Rear Yate Loons. In the case of the Hyper 212+ and H50, the rear is the Blademaster, otherwise all fans are Yate Loon Mediums except Top Rear is NZXT Gamma stock fan (except in case of NH-D14). Basically, I have a ton of fans pulling air into the case, and there very good air flow. Cleaned.
I also did quite a few remounts of each paste, different configurations, but let's say for evidence's sake, I did a single mount of each heatsink. I believe my results are within 1*C but I'd say 2*C error to be more accurate.
Listed is the max temp (which consistently was 0.5*C above median temp) as noted by HWInfo64.
Thermal Paste Comparison
Warning: Spoiler! (Click to show)
Warning: Spoiler! (Click to show)
PK-1 and PK-2 perform the same, As just an update to PK-1, PK-2 is definitely better and simply easier to spread. I did not find PK-1 that hard to spread though.
Warning: Spoiler! (Click to show)
PK-3 is way better. This revealed itself in multiple reseats too, it was quite surprising. Given that we all know that PK-1 is arguably one of the best pastes out there, PK-3 just blows everything else away. There have been a few new pastes recently - Masscool, Gelid GC Extreme, and Phobya Hegrease i think, that are quite high end, but I don't see them being significantly better than PK-3 given that PK-3 is significantly better than PK-1. Just imagine if this these tests were conducted on an Intel set-up where the temperature ranges are in the 80's, you'd probably see even bigger differences in different thermal pastes and coolers there.
For coolers, the Hyper 212+ is actually pretty terrible and more like $20 to make it worth it, than $20 is such a great value. Midrange cooling options at $40-50, like a Corsair H50, is good value (and at cheaper, a great value), and offer a small, but significant, upgrade to the Hyper 212+ . The real value, truthfully, is in high class coolers like the NH-D14 on sale for under $60 (for $50 or less, it's just a steal), and definitely a better value than the Hyper 212+ or other low end coolers at $20 or less, or mid-range coolers for $40 or less. Frankly, at $70+, they aren't priced too terribly, although I'd say $80+ is a bit too much.
The best cooler to buy though, is what you need, but for any overclocking chip, you'll probably need at least mid-range cooling, and you might need high end cooling if your pushing the limits of a 24/7 overclock or got a hot chip, or just an overclocking enthusiast.
The Corsair H50 is definitely an upgrade to the Hyper 212+. In fact, the Hyper 212+ it not really that good a cooler, it's simply okay for the price, but given that you can find the Corsair H50 for $20 used on ebay, which is as cheap as a brand new Hyper 212+ on sale or after rebate, I think it's the better buy. You can also find all sorts of closed loop 120mm radiators going for insanely cheap these days, for some reason they are all on sale around $30-50, at which price they are a better buy than the Hyper 212s for value, especially if you put better fans on them.
The H50 was tested not just to show the performance of closed loops though, it's also to compare the Hyper 212/NH-D14 (ie low and high end cooling) to any mid-range cooler. The Corsair H50 performs in dead heat, generally, to Prolimatech Megahalems, TRUE 120 & Venomous X, and other popular mid-range coolers. At above $40, I don't think mid-range cooling is worth it, even higher end mid-range coolers like the Antec 620 or H60, but if you can find such a cooler on sale, you might want to consider it's value over the Hyper 212, or any low end cooler in general.
Finally, the NH-D14. Wow, what a value. With an 11*C temp drop, just at the 50's, you can see why dual towers tend to be around $70-80. However, I found my NH-D14 for $43 (after shipping) on Ebay, and sometimes you can find dual towers like the Assassin and NH-D14 for $39-59 if you find the right deal, or used.
If you can find a dual tower under $60, it's simply a better value than the Hyper 212+ at $20. I was truly shocked at how much of a performance gain is to be had with a high end heatsink, and I really expected that a high end air cooler wouldn't be much better, and just a ridiculous price premium for a couple degrees, but I am pleasantly surprised to be completely wrong on that.
The Hyper 212+ simply is not adequate cooling for overclocking an overclocking CPU. It gives a large temp drop over stock cooling, sure, and at $20 it's quite cheap, but if you plan to push the 24/7 voltage limits of your chip, or for higher end overclocks, it won't be enough. Given that you can find other coolers at great deals, I think it's best to look past the Hyper 212+ for another cooler these days. I don't think I'll be recommending the Hyper 212+ to fellow overclockers.
On the other hand, if you just need just a few more degrees, consider buying some new thermal paste. You can find PK-3 @ 1.5g for under $4 at some places, including shipping. Now after this review, I do believe PK-3 is going to be the best conventional paste out there (non-destructive, standard application paste), considering how well just PK-1 compares to most pastes out there. But Phobya Hegrease (more expensive than most pastes), Coollaboratory Liquids, IC Diamond, even at $15-25, for 2-10*C price drops from most other pastes, is still a good deal.
All in all, I've been surprised by the results. If anyone is out there considering buying aftermarket cooling, seriously look into higher end solutions, as well as better fans, or thermal paste, for extra degrees. You can get a significant decrease in temps with higher end cooling, and you can also get a significant drop in temps by switching to a better thermal paste. I did not expect such a significant change by just changing my thermal paste or cooler, as I did in these tests.
I'd also like to thank Prolimatech for giving me the opportunity to do this. They were quite helpful and gave me a lot of feedback with some questions I had in doing all of this.
This is just general impressions of products, my thoughts on their build and such, overall quality, and pricing. I hate when reviewers talk about how great a cooler is for the price, yet at the same time I hate when reviewers bag on a product, like a $30 PSU for not being modular, because they don't account for it's price. The best reviews, are the ones that talk about absolute quality, features, while being aware of the item's price, and discussing at what price limit the product is at (ie what is the limit at which the product is priced fairly, what price it's a steal, what price it should be).
Warning: Spoiler! (Click to show)
PK-3 is awesome. I was surprised by how much better it was than PK1, which was significantly better than my stock CM paste. You read that improved paste will only give you like a degree or two at most, but I was consistently getting 3-5*C just from PK3 vs PK1 over multiple remounts. I was also getting consistent results from reseating pastes over and over. How you spread largely doesn't matter much, it's more about quantity. Just don't use the spread method (even though Prolimatech recommends it with their paste), it gave me disastrous results every time I did it.
You also don't need to cover the entire IHS, I was surprised by that. As in, consistently, having paste only cover around the die was more than enough, and temps, if anything, dropped in my attempts to put enough paste on so it would cover the entire IHS. You really can't use too little TIM, a small dot or grain of rice is more than enough, you shouldn't use more than that. If you use enough paste to cover the IHS you are probably using too much.
At $4 for 1.5g of PK-3 on ebay, it's a great way to shave a good 3-8*C off (depending what you currently use), which is a price/performance ratio that's hard to find in aftermarket cooling, you only get that like compared to the stock cooler.
Also, although I didn't get to do authoritative tests on the subject, this stuff outclassed AS5. I don't know why anyone recommends such a terrible paste still, it's terribly outdated by now. There's so many better pastes for the same price, and it's hardly proven or convenient when it's the most inconvenient paste around. Modern pastes are generally focused on being convenient, less conductive, less toxic, less viscous, less corrossive, no burn in time, easier to remove, as pastes these days are all really good performers. While I don't really care about these kinds of things, given that modern pastes are so focused on this, I do have to note that PK1, PK2, and PK3 didn't appear to have any burn in time at all. They all seemed pretty easy to deal with, although I could tell PK2 was easier than PK3.
This was my first cooler, and a great entry-level heatsink, of course. It is the most popular heatsink out there, too.
Mounting with the Hyper 212+ is fairly easy. The Heatpipe Direct Touch design makes the heatsink very cheap, and perform very well at the lower temp ranges or powers, but will not stand up to hotter temp ranges. This cooler was all I needed for an Athlon II build, but with a Phenom II it was really stretching it, and I needed performance paste, a 2nd fan, tons of case fans, just to push 3.7ghz. Now for gaming and general usage, and even in Prime95 blend testing, I'm not going to see temps nearly as high as Small FFT, but with encoding and streaming, as I do, you still get hot temps and you'd rather not push the limits in general encoding usage.
A great first heatsink, and recently around $19, it's decently priced, but any sort of K series or black edition or high end overclock will need more than a hyper 212+.
The paste it comes with isn't half bad, but is not a performance paste. I've found the best way to paste this thing, is to use the line method, as outlined by BenchmarkReviews. You might be tempted to use the spread method on this kind of heatsink - don't. You'll end up using an entire bottle to fill in the cracks, you'll just wipe the paste off the raised areas. Just a few lines on the heatpipes, and the rest of the paste will spread out. Don't use too much.
A 2nd fan will benefit the heatsink by a degree or three. The heatsink feels solid, the mount is solid (but will allow you to wiggle the heatsink around, for better or worse). It's fairly easy to mount the heatsink and to mount/remount/unmount it, as well as take off the fans.
At more than $20, I'd say this heatsink is not worth it. I wouldn't pay more than $25 for the Hyper 212 Evo, either, or any other low end cooler. It's a much better value than most of it's peers, for sure, but nowadays it is getting a little outdated and outpaced, and you should be able to find a better value cooler or a better cooler with equal value, and if you're spending only $20 frankly I think you are better off just buying a used or flawed mid-range cooler off ebay, as you don't have much to lose with just $20 (it's not like a Hyper 212+ has any resale value...).
I was pretty impressed with this heatsink. I came across the H50 because I was looking for a way to upgrade from the Hyper 212+, and pay for a good value. In essence, I was looking for a Hyper 212+ that was a step up from the Hyper 212+. You hear everyone recommend the Hyper 212+, but you don't really hear the same clamor for mid-range coolers, and you don't hear the words "value" when it comes to high end coolers exactly either (although it does exist).
For some odd reason, the H50 tends to sell at really discounted used prices. Now, you might say that's a dumb observation, don't all heatsinks, and all things in general, sell cheaper used? No - what I mean, is that the Corsair H50 sells for much cheaper used. You can find Corsair H50's around 50-70% cheaper used than new, while most heatsinks tend to sell only around 20-30% cheaper when used, even very similar closed loops. There's a few other coolers that tend to sell very cheap when used as well (True 120, which is old as bones now, Zalman 9900Max, which is odd and can't mount different fans, Zerotherm ZT-10D which no one has heard of), but they are all slightly more expensive for the same performance or a bit worse for only slightly cheaper.
And, if you find a 'defective' Corsair H50, you can just find these for a handful of dollars. Literally. Corsair will send you a free mounting kit if you email them, but a Corsair used, with stripped screws (which cost pennies at the hardware store, by the way...), will run about $11. I found my Corsair H50 for $27 after shipping.
And, after Christmas, prices have just plummeted for Corsair H50's. Paying over $30 (including shipping) nowadays for a used Corsair H50 is too much. While you might question buying something used, the performance on these is so good that it's worth it, and you can't go wrong with $20 for a closed loop water cooler.
Mounting the Corsair H50 is a bit of a pain. Online, you hear a lot of people recommend the Corsair H50, because as a closed loop it must be easy to install. Not so - I had a much harder time making the Corsair H50 fit, than my NH-D14 (which I only had to remove a custom VRM fan to fit, it still fit with my custom VRM heatsink, and the Nh-D14 cools your VRM too with it's 140mm).
The radiator on the H50 is slightly taller than a 120mm fan, so don't think it'll simply fit anywhere a 120mm fan would. I could not mount this fan on any 120mm fan location on my case because of this - there are small, ~1cm additions both on top and bottom of the radiator where the water runs through. If you want to use push/pull, which should be a requirement for this thing (more on that later), you're taking up 25mm + 25mm + 25mm. In fact, I believe the H50 is slightly more than 25mm, it'd be more like 30mm.
80mm comes out pretty far, as evidenced in the picture section. It will be right on top of your CPU in a mid-size ATX case, which makes things awkwards with the hoses (a whole thing on it's own...). If you have spot cooling or custom cooling on your VRM, you might have compability issues. Basically, if your VRM is tall like tall RAM, you might have an issue. Also, this may prevent you from using top fans, because of how the radiator is 120mm + some.
The tubes are extremely stiff on the H50 - this has been changed with the updated H55 and the newer H60/H80 models. The corrugated tubing is meant to be extremely evaporation resistant, so the unit will last forever and the water in it won't slowly dissapear over time (which does happen). I'm sure this is all true, but that doesn't mean they aren't any less of a pain in the behind to deal with. You'll find yourself fighting the tubes, and they are so stiff you feel like it'll snap when you try to mount the radiator or waterblock, and the tension is so tight the radiator will swing like a metal bowling pin in a cartoon (and crash through your case or break your face...).
I loved the Corsair H50, but one thing it is not, is convenient. Don't let this deter you from buying this cooler, it's a great cooler, but it is NOT a convenient cooler! It's not a simple "Oh well since it's water I'll just mount the radiator anywhere!" There's a limited turning radius on the tubes, and Corsair even recommends you mount the radiator to reduce 'hose torque' (the pressure can be so great from the stiff wires that it'll actually affect the waterblock mount on your CPU).
If you want the Corsair logo to be pretty and level, you're gonna have a very hard time mounting this cooler somewhere. The hoses will dictate how the logo will be oriented, basically. There is discussion that the Corsair H50 may or may not be better mounted one way or another, and in regards to it's orientation, as well. I found that it's perfectly fine for the radiator to be mounted vertically, up top or on the rear, and intake vs outtake. What is important though, is that the cooler gets cool air, so as long as you have good case air flow, it's okay for it to be exhaust. I did have different results in where the cooler was mounted, and it was generally based on if it was getting good air flow or not, not how it was positioned.
What is important with the H50 though, is fans. Unlike the Hyper 212 and NH-D14, the radiator gets hot! Normally, when you touch an air cooler, it doesn't feel warm, even on load. It might even feel cold, although usually it'll feel just ... 'bleh' (you know, how metal absorbs heat and so when you touch metal that's room temperature, it'll feel cold, because it's sucking heat off of you, basically). Well, the Corsair H50 radiator definitely is very warm on load. Obviously, the water at work, and the H50 is limited by the radiator size and amount of water in the system. Quite simply, the stronger the fans you use, the better.
I imagine an H50 with a reservoir and radiator mod would drastically reduce temps (as evidenced by people who've done it).
I did a small test even, where I used the loud and powerful Blademaster instead of a Yate Loon Medium as Pull, and got a 3*C temp drop instantly. I'm sure the sky is the limit, and if you put, say, 2 x 140mm fans on this thing, it'll cool almost as well as high end air. I'd actually recommend for anyone using an H50, to do that.
The water block makes a sound, but it is very quiet. It's sort of a low, grinding sound, but it's not annoying at all and you'd have to stick your head in the case, just to barely hear it. Also, the H50 supposedly does not have as many issues with leakage as some other closed loops may have. You can always doublewrap it or something to prevent leakage anyways, I wouldn't worry about it.
The mounting system on the H50 is a little barbaric. In order to mount it, you have to screw in a metal clamp thing, but you can't screw it in all the way, and then you stick in the unwieldy waterblock, twist it, and then screw it all the way. All the while the metal clamp thing is loose, and you'll have to pull it up, or else gravity will make it go flat against the board and the rear retention plate will just sort of fall off... hard to explain.
It's not a big deal, I did plenty of remounts on the H50, but of the hyper 212+, NH-D14, stock cooler, and h50, it was the most difficult to deal with, when ironically it's supposed to be the easiest. This shouldn't dissuade you from buying the cooler at all, and it's not like it'll be hard to mount if you're new to building computers or anything. It's not like it's difficult to install, it's just unwieldy. Like doing it with a fat girl, you know where things go, it just takes a little more effort.
I was ultimately quite impressed with the cooling performance I got out of this thing, it's definitely an upgrade from a low end cooler like the Hyper 212+, but it's not a huge upgrade. If you need a slight step up, or just a few more degrees, from your low end cooler, ten this is a great step up, especially if you combine it with some improved paste and fans. You will also need at least 2 fans, if not 2 high power fans, to make this thing work for you, but going from a Hyper 212+, to an H50 with PK-3 and 2xYate Loon highs would probably give you a significant bump up if your Hyper 212+ was just barely getting too hot for your overclock.
I'm sure the H55/H60/Kuhler 620, with smooth tubing, is much easier to work with though, but I wouldn't pay more than $40 for a mid-range cooler in general (I believe some, like the 620, is on sale for $39 right now though).
The stock fans with all closed loops suck, and don't be fooled into 'closed loop is quiet'. You need powerful fans to dissipate the large amount of heat the water absorbs, it's not going to be any quieter than a tower (or much quieter, that is). If you are looking for quiet, look into something like an HR-02 Macho, or big block cooler (dual towers tend to make a lot more noise compared to single tower, and I mean with identical fan set-ups, and single tower runs better than dual on less/lower fans). My point is the H70 is crap, since it's worse than the H50 in apples to apples, but comes with 2 crap fans that won't be powerful enough, since you'll be buying your own fans for closed loop coolers anyways.
The point of the Corsair H50 test though, is to show the performance increase of any similar mid-range cooler over a low end cooler like the Hyper 212 - a True 120, megahalems, mugen, etc, are all worthy heatsinks are $40-50 or lower, and are not if they are more than that, give or take based on individual performance.
Finally, be careful mounting this thing. While you will scratch up any heatsink base if you try to touch it to metal without paste (like trying to see how it would contact the cpu before pasting...), with how tense the hoses are and how it mounts, if you don't keep a strong hold of the water block at all times during the mounting process, it'll just fling up and scrape on something or break someone's nose or your motherboard.
I really didn't think a high end cooler would do much, but I have to say that this dual tower, at $43 on ebay, was a better value than the Hyper 212+ at $19.99 I didn't quite expect such a significant decrease in temps, I was getting almost 15-20*C in some test configurations on just a Phenom II system. I guess if you look at bad reviews that test stock voltages, you wouldn't know these kinds of things though (cough frostytech/hardwaresecrets/th cough).
The heatsink is quite huge. I love how it looks, but then again I think all coolers like pretty rad, from the Hyper 212 to the NH-D14. The bigger , the cooler.. The base appears to even be pre-lapped, as it has a nice finish and glean to it (not quite mirror, there is a machining pattern, but it does have a reflection). It is ridiculously heavy though, and will make your case feel like a sack of bricks.
Mounting this heatsink is pretty simple, but you'll need a long screwdriver, since the screws are in between the 2 towers. Make sure your screwdriver is longer than the heatsink is tall. It also requires a bit of pressure to have the screw make contact. It's not as easy to install/uninstall as the Hyper 212+, but it isn't exactly difficult either. The fans also can be adjusted and raised/lowered for compatibility.
I'd strongly recommend any dual tower if you can afford one, under $60. You can usually find a NH-D14/K2/Assassin for sale somewhere or used for under $60, at which point they become better values than lower end and 'budget' coolers. Given the value of heatsinks, I'd say these are the true budget coolers. You might not need the performance of a dual tower, but if you do, don't hesitate to get one for under $60.
The fans are pretty quiet on this thing. I actually bought this as a NH-D14 SE2011 missing a mounting kit for my AM3 system for $43, and got a replacement 3rd fan mount, sticker, AM3 and LGA1155 mount, and pwm y spplitter, from noctua for free. The price was so low because it was missing a mounting kit, but Noctua gives you free mounting kits anyways. It's also noice that the SE2011 comes with PWM fans instead of regular fans. It's the same heatsink but comes with different mount at stock, and PWM fans.
Tight Fit, H50 Rear Mount Push/pull. You can see how the radiator isn't exactly 120x120.
Top Mount. For science.
This is how I ran all comparison tests. Temps were maybe 0.5*C best at top mount, but inconclusive evidence.
Hyper 212+ Spread Method. This was like half a tube, what a jip. Don't use spread method.
And this is spread method covering entire IHS. I dropped temps by another 5*C by simply taking a towel, and a single wipe over the base to get most of this crap off. As you can see the base of the heatsink in a few spots, that just tells you how thin this application was too.
PK-1 Spread. This is around the optimal amount, although it will only over 1/2 to 2/3rd of the IHS, which is perfectly fine. You want to cool right above the die, not the entire IHS, as the heat doesn't dissipate across the entire IHS
PK-3 spread. Despite the most minimal coverage, the most significant results, including over other mounts
PK-2 Remount. Identical results to previous mount, just making sure.
NH-D14 PK-1. 2 Rice grains perpendicular to mounting pressure (2 screws on nh-d14 mount), is full spread across IHS. No difference than single rice grain...
NH-D14. This is what way too much paste looks like before mounting
. This ended up being so much paste that is caused about 4-5*C increase in temps over a proper, single rice grain method (or even double rice grain method). So keep this in mind next time you apply your paste and think you need a little more. It's very easy to use too much paste, but it's nearly impossible to use too little.