It would be nice to see some common heatsinks within the same price bracket included in the comparison charts, even if it's only a handful. They're direct competition after all.
I have to agree, otherwise the results are quite relative. Put in at least a couple for reference (like Intel highend stock, AMD highend stock, and some elite one like Noctua NH-D15), it would make the info much more worthwhile for general public
Sometimes I wonder... if some of these reviews are sponsored (either by one manufacturer or a group of manufacturers, to boost brand exposure and potentially sales)... and not necessarily direct monetary sponsorship. I also wonder... if the reason Anandtech didn't include a high-end air cooler (e.g., NH-D15)... is because it would make a strong case for the AIO solutions making a very poor value proposition... especially when considering noise/cooling ratios... and, in turn, rendering this entire review pointless.
Actually, the answer to your question is far less complicated than that. We only compare air coolers to air coolers and liquid coolers to liquid coolers because each represents a specific type of product. It's an "apples to apples" kind of thing. I also personally do not believe that these two actually compete with each other. Users are much more likely to go for an AIO cooler because they want easier access to their system, or because they just don't like a 2kg piece of metal hanging off their motherboard, or just because it "looks cool", than for the performance. If acceptable performance is all that a user needs, an average air cooler can do the job just fine, they do not even need a NH-D15 to begin with.
Besides that, the magic of having and using professional equipment like ours is that everything is both repetitive and comparable. They were all tested on the exact same device and everything is in our methodology page that can be found in every cooler review. Long story short, you can easily go to our review of the NH-D15 cooler (or any cooler that you'd like to compare, regardless of its type and platform) and just compare any of its performance figures to any AIO cooler that you want to, taking into account your personal requirements too.
Sorry no.. it is green apples to red apples in most buyers considerations.. or dreams, and they really need to know what they are paying extra in case their dream comes true.
I think that "apples to apples" does not hold here. After all, both liquid and air coolers serve the same purpose. It would be nice to be able to compare them without having to do research through Anandtech's archives. Is it possible that once comparison is made obvious liquid coolers don't make sense from the cost/performance point of view? I don't know since review does not try to inform me from that perspective. My guess is that there are many people who would like to inform themselves of whether upgrade to a liquid cooler would make sense or not but to do that you recommend that they dig though the past reviews instead of you offering that information in your review. All for the sake of keeping apples and oranges separated. I don't buy that reasoning.
By the time you whined about how things should be done, I had two pages opened, I can tell you this: you get lower noise on air coolers in most scenario but delta over ambient on the worst AIO is lower than the best air cooler at 150w load.
I would have loved it if there was two relevant air coolers included in these graphs but nothing to write a page about.
You are asking for apples to apples comparisons on coolers with requirements *that aren't the same* (packaging is a huge deal), therefore your logic breaks down.
I agree with what tricomp and rtfmx9 said... the "apples to apples" reasoning doesn't apply here. From personal experience a ton of people are always asking is water cooling better than air. From a functional standpoint they both serve the same exact purpose.
I buy AIO coolers mainly because I don't care about getting max cooling efficiency and I think AIO coolers just look far far better than chunks of chrome plated tin in my rig. I'm sure I'm not the only one.
I'm sure you're not. I personally have a side panel on my case and can't see the cooler, so effectiveness and price (value) are the factors important to me.
I am so glad that you are here to tell your audience that they are wrong and they don't actually want the products compared, despite the numerous comments suggesting and requesting a direct comparison. You must work for Apple. Even though we all ask for the comparison, and believe that they are competing with each other (after all, they serve the same purpose, cooling your system, so I fail to see how you can't correlate that they would be competing for our dollars), yet, we should just forget our own thoughts, and jump on your bandwagon of no need to compare in the review because they are entirely different types of cooling...
Then you finish it off with your condescending "the magic of our professional equipment" line, and "you can easily go look up previous reviews the the information you want" attitude. You are the one being payed to write these reviews. Yes a ton of use have used the past data to find things, but it isn't always so clear cut, if you can't remember an exact model name, or as has happened, the data isn't on the site anymore, or is a bear to find.
Seriously, in the past, I remember they used to have a drop down menu with all the reviewed GPU, or CPU and it was easy to find the information. I don't see that anymore, since the site got redesigned.
So get off your high horse, and lose the snarky attitude. I swear, since Purch bought Anandtech the quality and the attitude has stunk.
Ok, I found the the area I was referencing. "Bench" at the top of the site and yes, the cooling data is in there. Though, I would hope the methodology has remained constant.
Air coolers and liquid coolers absolutely are direct competition with one another and I think it's incorrect to state that they're not. It'd be helpful to a number of readers to include a couple of air coolers (maybe the OEM sorts to form a baseline and one or two aftermarket models) without telling your readers to go fish in old articles. :(
A rather disappointing comment, especially coming from AT. You claim testing is repetitive and that we can look at (much) older reviews and trust the numbers are comparable.
While you are not exactly an engineer, you could document yourselves a little better. The ambient temperature could not have been exactly the same. You persist in showing delta above ambient in all your reviews, which is an incredibly misleading figure for anyone without an engineering background: temperature does not have a linear dependency with the causing factors (e.g. power draw increases quadratically with frequency, etc) and the same goes for ambient temperature. Heat up your room by 20C and then run the exact same tests and you'll see different delta-over-ambient figures.
You should stop showing delta-over-ambient and instead include ambient temp and device temp in each of your graph. That's far more informative.
You don't think your readers want to know if a cheaper and more standard fan/heatsink cooler works better than one of these bulky AIO units? Seriously?
I have never used an air cooler that can work as well as an AIO cooler. Sure it would be nice to see them compared, but a liquid cooler can handle a MUCH higher load than a heatsink and fan can. And I have used the same CPUs with both style coolers. I end up with a MUCH cooler and quieter setup with the AIO cooler.
"The RGB strip is a welcome addition that surely every user with a windowed side panel will appreciate."
The only part about the RGB Cancer Strip that I'd appreciate if I owned this is that it's not hard wired in, meaning that chemotherapy will only consist of tossing the strip in my to recycle box and not require voiding the warranty or having some piece of garbage grade software run at startup to turn them off each time I boot my system.
Great write up & lots of info on the product but I have to ask this. When doing a test of a cooler and only giving the Delta type stats like here & GN does the same thing & I also said this to them as well. What is the point of doing a cooler temp test if we do not get the actual temps of the cooler under the tested work loads? Doing a Delta over ambient means very little if we do not know the ambient temp at the time for each of the testing runs done. I think we also need idle and max temps put into the charts so we know exactly how each cooler does against each other and how hot a CPU will run at each tested work load. I know my own CPU overclocked can go as high as 78c in the stress tests but only reach 56c while gaming.
It makes a lot more sense to give delta temps. Let's say delta is 50c, you could have an ambient of 10c and a load temp of 60c, or an ambient of 30c and a load temp of 80c. Both are equivalent, and show the cooler being able to reduce temps from a given heat source to 50c above ambient. Why would you want to know the actual load temp and ambient?
I think it would be nice to have both sets of data to be included in the results as well as it clearly stating the ambient temp posted for each of the tests. This way everyone gets to see all the info they need. Personally I also like to see actual CPU's being tested so it puts actual models of cpu's tested and each of them well behave with each cooler. With how companies like how Intel use cheap paste under the lid and if you want to get the best temps you need to delid a Intel cpu to get better results testing on a heat block does not really show how todays Intel cpu's will behave with a cooler.Well that is if you do not do the delid of coarse.
With my own CPU which is a Sandy bridge soldered heat spreader i7 2700K I know how it will behave with any given cooler tested because of it's better build quality than their newer stuff. I know with my cooling setup which is air cooler that when I had it in the basement @ 19c room temp it does 5.0GHz-5.1GHz and stays at pretty good temps under normal work & gaming loads.
I have since moved that system upstairs and room temp is 21c-22c it will do 4.9GHz-5GHz because the temp went up a bit under normal work and gaming loads by a few degrees. Not enough that it became unstable or over heating but just enough to make me uneasy since it would hit 84c-87c in the stress testers I use to test if it's stable and as to how warm it gets under those stress tests.
I want to switch to a big rad water cooling setup soon which is why I read or watch reviews for this now. By not showing actual temps with real cpu tested data it makes it almost useless for me to use this data to help me make my buying plans on which ever water cooler is the best. As you said it does give a delta temp over ambient and that you can suck the info from that to give a idea of how coolers will react to a given amount of wattage and heat they need to be able to remove to keep things chill.
The testing done here is done very well but it only paints part of the picture and needs the data I said needed to be added to make it more useful to everyone.
You have done a great job to eliminate “randomness” from your results, creating a testbed that allows for reproducible “objective” results.
But I wonder if that is entirely a good thing, because in practice these things *will* be running on CPUs, that are more dynamic in terms of heat generation with every generation: Just imagine the engineers finally come up with a non-volatile CPU design and energy consumption jumps between 0 and 200 Watts in microseconds.
Because humans used to be hunted (and hunt), our brain is designed to filter out constant noises, but be especially sensitive to *changes* in sound. So, if your aim is to hide the sound your PC produces, as independently from the current workload, you need to design not necessarily for the lowest absolute noise for any workload, but for *unnoticeability*.
The general problem is, that a jump between 10 and 800 Watts between in ideal idle desktop and a maxed-out gamer rig probably takes attaching a submarine (submerged) to cool without any noticeable acoustic signature: As you eliminate all moving parts you find that there are non-moving parts that are in fact creating sound as well, when having to deal with these power deltas. And I guess the ultimate way to counter these is going active, using noise cancellation techniques.
Now all that is probably excessive, but what I would recommend you to test is to see how these cooling rigs react to dynamic changes of the heat-supply: If their sound profile remains very steady under dynamic workloads, they may be far more acceptable than if they make frequent tiny adjustments even at very low absolute levels. That’s because our brain is tuned to amplify the changes and tune out the steady drones.
I have an NHD-15 heat sink and I also have a corsair h80i v2. (my case, a silverstone FT1 does not easily support radiators larget than 120mm) My CPU is an older 3930K six core that warms up nicely... Of course I do not have all the nice test bench stuff that Anandtech has, so my results are quite a bit more touchy-feely. By just a whisker the NHD-15 is both a tad quieter and a tad cooler.... I also like the fact that I know that outside of the fans there are no moving parts to fail.
I am sure that my limitations are the 120 mm radiator, and I am also pretty well convinced that within a given sized radiator, there will be a very narrow range of test results across the range of AIO kits. But I keep looking...
Since the advent of AIO liquid cooling kits my personal builds have changed a good bit. Have a H100i on my main build for a few years now, and my old and trusty second machine has a Hyper 212 EVO on it.
I've had bad experiences with heavy air coolers in the past, so I tend to stay away from them these days. My cases have positive pressure and the H100i works well as an exhaust mounted at the top.
To each their own. I know several people that frequent this website are still fond of air cooling, and in many situations it's still a very good and efficient option but in the end I'd say it depends on what kind of build you have in mind.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
34 Comments
Back to Article
Yuriman - Thursday, December 14, 2017 - link
It would be nice to see some common heatsinks within the same price bracket included in the comparison charts, even if it's only a handful. They're direct competition after all.HollyDOL - Thursday, December 14, 2017 - link
I have to agree, otherwise the results are quite relative.Put in at least a couple for reference (like Intel highend stock, AMD highend stock, and some elite one like Noctua NH-D15), it would make the info much more worthwhile for general public
HollyDOL - Thursday, December 14, 2017 - link
not like stock coolers could hope to do anything except bbq with 340W load ofc...Dr. Swag - Thursday, December 14, 2017 - link
Awesome! A decent alternative to the S24 when it comes to quiet AIOs that also has rgb!sweetca - Thursday, December 14, 2017 - link
Catchy name.Lord of the Bored - Saturday, December 16, 2017 - link
"I am Captain DeepCool, of the Federation starship EX RGB."ReeZun - Thursday, December 14, 2017 - link
Sometimes I wonder... if some of these reviews are sponsored (either by one manufacturer or a group of manufacturers, to boost brand exposure and potentially sales)... and not necessarily direct monetary sponsorship. I also wonder... if the reason Anandtech didn't include a high-end air cooler (e.g., NH-D15)... is because it would make a strong case for the AIO solutions making a very poor value proposition... especially when considering noise/cooling ratios... and, in turn, rendering this entire review pointless.E.Fyll - Thursday, December 14, 2017 - link
Actually, the answer to your question is far less complicated than that. We only compare air coolers to air coolers and liquid coolers to liquid coolers because each represents a specific type of product. It's an "apples to apples" kind of thing. I also personally do not believe that these two actually compete with each other. Users are much more likely to go for an AIO cooler because they want easier access to their system, or because they just don't like a 2kg piece of metal hanging off their motherboard, or just because it "looks cool", than for the performance. If acceptable performance is all that a user needs, an average air cooler can do the job just fine, they do not even need a NH-D15 to begin with.Besides that, the magic of having and using professional equipment like ours is that everything is both repetitive and comparable. They were all tested on the exact same device and everything is in our methodology page that can be found in every cooler review. Long story short, you can easily go to our review of the NH-D15 cooler (or any cooler that you'd like to compare, regardless of its type and platform) and just compare any of its performance figures to any AIO cooler that you want to, taking into account your personal requirements too.
tricomp - Thursday, December 14, 2017 - link
Sorry no.. it is green apples to red apples in most buyers considerations.. or dreams, and they really need to know what they are paying extra in case their dream comes true.rtfmx9 - Thursday, December 14, 2017 - link
I think that "apples to apples" does not hold here. After all, both liquid and air coolers serve the same purpose. It would be nice to be able to compare them without having to do research through Anandtech's archives. Is it possible that once comparison is made obvious liquid coolers don't make sense from the cost/performance point of view? I don't know since review does not try to inform me from that perspective. My guess is that there are many people who would like to inform themselves of whether upgrade to a liquid cooler would make sense or not but to do that you recommend that they dig though the past reviews instead of you offering that information in your review. All for the sake of keeping apples and oranges separated. I don't buy that reasoning.Galid - Thursday, December 14, 2017 - link
By the time you whined about how things should be done, I had two pages opened, I can tell you this: you get lower noise on air coolers in most scenario but delta over ambient on the worst AIO is lower than the best air cooler at 150w load.I would have loved it if there was two relevant air coolers included in these graphs but nothing to write a page about.
bananaforscale - Wednesday, June 20, 2018 - link
You are asking for apples to apples comparisons on coolers with requirements *that aren't the same* (packaging is a huge deal), therefore your logic breaks down.kevbev89 - Thursday, December 14, 2017 - link
I agree with what tricomp and rtfmx9 said... the "apples to apples" reasoning doesn't apply here. From personal experience a ton of people are always asking is water cooling better than air. From a functional standpoint they both serve the same exact purpose.jabber - Thursday, December 14, 2017 - link
I buy AIO coolers mainly because I don't care about getting max cooling efficiency and I think AIO coolers just look far far better than chunks of chrome plated tin in my rig. I'm sure I'm not the only one.Yuriman - Friday, December 15, 2017 - link
I'm sure you're not. I personally have a side panel on my case and can't see the cooler, so effectiveness and price (value) are the factors important to me.SilthDraeth - Thursday, December 14, 2017 - link
I am so glad that you are here to tell your audience that they are wrong and they don't actually want the products compared, despite the numerous comments suggesting and requesting a direct comparison. You must work for Apple. Even though we all ask for the comparison, and believe that they are competing with each other (after all, they serve the same purpose, cooling your system, so I fail to see how you can't correlate that they would be competing for our dollars), yet, we should just forget our own thoughts, and jump on your bandwagon of no need to compare in the review because they are entirely different types of cooling...Then you finish it off with your condescending "the magic of our professional equipment" line, and "you can easily go look up previous reviews the the information you want" attitude. You are the one being payed to write these reviews. Yes a ton of use have used the past data to find things, but it isn't always so clear cut, if you can't remember an exact model name, or as has happened, the data isn't on the site anymore, or is a bear to find.
Seriously, in the past, I remember they used to have a drop down menu with all the reviewed GPU, or CPU and it was easy to find the information. I don't see that anymore, since the site got redesigned.
So get off your high horse, and lose the snarky attitude. I swear, since Purch bought Anandtech the quality and the attitude has stunk.
SilthDraeth - Thursday, December 14, 2017 - link
Ok, I found the the area I was referencing. "Bench" at the top of the site and yes, the cooling data is in there. Though, I would hope the methodology has remained constant.DanNeely - Thursday, December 14, 2017 - link
It probably is. For other tests when methodology changes the benchmark name changes; which is why you've got things like "GPU 2017".PeachNCream - Thursday, December 14, 2017 - link
Air coolers and liquid coolers absolutely are direct competition with one another and I think it's incorrect to state that they're not. It'd be helpful to a number of readers to include a couple of air coolers (maybe the OEM sorts to form a baseline and one or two aftermarket models) without telling your readers to go fish in old articles. :(normadize - Friday, December 15, 2017 - link
A rather disappointing comment, especially coming from AT. You claim testing is repetitive and that we can look at (much) older reviews and trust the numbers are comparable.While you are not exactly an engineer, you could document yourselves a little better. The ambient temperature could not have been exactly the same. You persist in showing delta above ambient in all your reviews, which is an incredibly misleading figure for anyone without an engineering background: temperature does not have a linear dependency with the causing factors (e.g. power draw increases quadratically with frequency, etc) and the same goes for ambient temperature. Heat up your room by 20C and then run the exact same tests and you'll see different delta-over-ambient figures.
You should stop showing delta-over-ambient and instead include ambient temp and device temp in each of your graph. That's far more informative.
IdBuRnS - Wednesday, January 3, 2018 - link
You don't think your readers want to know if a cheaper and more standard fan/heatsink cooler works better than one of these bulky AIO units? Seriously?Stuka87 - Monday, December 18, 2017 - link
I have never used an air cooler that can work as well as an AIO cooler. Sure it would be nice to see them compared, but a liquid cooler can handle a MUCH higher load than a heatsink and fan can. And I have used the same CPUs with both style coolers. I end up with a MUCH cooler and quieter setup with the AIO cooler.ellis.bentley - Tuesday, January 9, 2018 - link
Great post you have shared with everyone. We can get valuable knowledge from this site. Thanks for sharing this informative stuff with us. https://stjohnsrcschool.org/apps-for-pc/imessage-o...https://stjohnsrcschool.org/apk/uktvnow-apk/
DanNeely - Thursday, December 14, 2017 - link
"The RGB strip is a welcome addition that surely every user with a windowed side panel will appreciate."The only part about the RGB Cancer Strip that I'd appreciate if I owned this is that it's not hard wired in, meaning that chemotherapy will only consist of tossing the strip in my to recycle box and not require voiding the warranty or having some piece of garbage grade software run at startup to turn them off each time I boot my system.
The_Assimilator - Thursday, December 14, 2017 - link
Amen brother!bananaforscale - Wednesday, June 20, 2018 - link
There's a non-RGB model that glows white.rocky12345 - Thursday, December 14, 2017 - link
Great write up & lots of info on the product but I have to ask this. When doing a test of a cooler and only giving the Delta type stats like here & GN does the same thing & I also said this to them as well.What is the point of doing a cooler temp test if we do not get the actual temps of the cooler under the tested work loads? Doing a Delta over ambient means very little if we do not know the ambient temp at the time for each of the testing runs done. I think we also need idle and max temps put into the charts so we know exactly how each cooler does against each other and how hot a CPU will run at each tested work load. I know my own CPU overclocked can go as high as 78c in the stress tests but only reach 56c while gaming.
Yuriman - Thursday, December 14, 2017 - link
It makes a lot more sense to give delta temps. Let's say delta is 50c, you could have an ambient of 10c and a load temp of 60c, or an ambient of 30c and a load temp of 80c. Both are equivalent, and show the cooler being able to reduce temps from a given heat source to 50c above ambient. Why would you want to know the actual load temp and ambient?rocky12345 - Saturday, December 16, 2017 - link
I think it would be nice to have both sets of data to be included in the results as well as it clearly stating the ambient temp posted for each of the tests. This way everyone gets to see all the info they need. Personally I also like to see actual CPU's being tested so it puts actual models of cpu's tested and each of them well behave with each cooler. With how companies like how Intel use cheap paste under the lid and if you want to get the best temps you need to delid a Intel cpu to get better results testing on a heat block does not really show how todays Intel cpu's will behave with a cooler.Well that is if you do not do the delid of coarse.With my own CPU which is a Sandy bridge soldered heat spreader i7 2700K I know how it will behave with any given cooler tested because of it's better build quality than their newer stuff. I know with my cooling setup which is air cooler that when I had it in the basement @ 19c room temp it does 5.0GHz-5.1GHz and stays at pretty good temps under normal work & gaming loads.
I have since moved that system upstairs and room temp is 21c-22c it will do 4.9GHz-5GHz because the temp went up a bit under normal work and gaming loads by a few degrees. Not enough that it became unstable or over heating but just enough to make me uneasy since it would hit 84c-87c in the stress testers I use to test if it's stable and as to how warm it gets under those stress tests.
I want to switch to a big rad water cooling setup soon which is why I read or watch reviews for this now. By not showing actual temps with real cpu tested data it makes it almost useless for me to use this data to help me make my buying plans on which ever water cooler is the best. As you said it does give a delta temp over ambient and that you can suck the info from that to give a idea of how coolers will react to a given amount of wattage and heat they need to be able to remove to keep things chill.
The testing done here is done very well but it only paints part of the picture and needs the data I said needed to be added to make it more useful to everyone.
abufrejoval - Thursday, December 14, 2017 - link
You have done a great job to eliminate “randomness” from your results, creating a testbed that allows for reproducible “objective” results.But I wonder if that is entirely a good thing, because in practice these things *will* be running on CPUs, that are more dynamic in terms of heat generation with every generation: Just imagine the engineers finally come up with a non-volatile CPU design and energy consumption jumps between 0 and 200 Watts in microseconds.
Because humans used to be hunted (and hunt), our brain is designed to filter out constant noises, but be especially sensitive to *changes* in sound. So, if your aim is to hide the sound your PC produces, as independently from the current workload, you need to design not necessarily for the lowest absolute noise for any workload, but for *unnoticeability*.
The general problem is, that a jump between 10 and 800 Watts between in ideal idle desktop and a maxed-out gamer rig probably takes attaching a submarine (submerged) to cool without any noticeable acoustic signature: As you eliminate all moving parts you find that there are non-moving parts that are in fact creating sound as well, when having to deal with these power deltas. And I guess the ultimate way to counter these is going active, using noise cancellation techniques.
Now all that is probably excessive, but what I would recommend you to test is to see how these cooling rigs react to dynamic changes of the heat-supply: If their sound profile remains very steady under dynamic workloads, they may be far more acceptable than if they make frequent tiny adjustments even at very low absolute levels. That’s because our brain is tuned to amplify the changes and tune out the steady drones.
croc - Friday, December 15, 2017 - link
I have an NHD-15 heat sink and I also have a corsair h80i v2. (my case, a silverstone FT1 does not easily support radiators larget than 120mm) My CPU is an older 3930K six core that warms up nicely... Of course I do not have all the nice test bench stuff that Anandtech has, so my results are quite a bit more touchy-feely. By just a whisker the NHD-15 is both a tad quieter and a tad cooler.... I also like the fact that I know that outside of the fans there are no moving parts to fail.I am sure that my limitations are the 120 mm radiator, and I am also pretty well convinced that within a given sized radiator, there will be a very narrow range of test results across the range of AIO kits. But I keep looking...
LordanSS - Friday, December 15, 2017 - link
Thank you, Mr. Fylladitakis for your review.Since the advent of AIO liquid cooling kits my personal builds have changed a good bit. Have a H100i on my main build for a few years now, and my old and trusty second machine has a Hyper 212 EVO on it.
I've had bad experiences with heavy air coolers in the past, so I tend to stay away from them these days. My cases have positive pressure and the H100i works well as an exhaust mounted at the top.
To each their own. I know several people that frequent this website are still fond of air cooling, and in many situations it's still a very good and efficient option but in the end I'd say it depends on what kind of build you have in mind.
snarfbot - Sunday, December 17, 2017 - link
how long are these coolers being run at load?deadlockedworld - Wednesday, April 11, 2018 - link
For all the commentary about the bench -- this deepcool is not in the Anandtech bench section. ...