POST A COMMENT

52 Comments

Back to Article

  • C'DaleRider - Tuesday, August 19, 2014 - link

    Looks like it's suffered the fate most radiators that used round tubes had....poor performance, and probably due to laminar flow issues. Read on another forum about testing done by some of the "pioneers" in watercooling and issues faced by round tube radiators. Lack of surface area being actively cooled and laminar flow problems.

    Ever wonder why the "real" water cooling crowd has almost no round tube radiators to choose from?
    Reply
  • liu_d - Tuesday, August 19, 2014 - link

    The reason most rads don't use round tubing is cost. If you look at the few radiators on the market that do (Aquacomputer Airplex, Feser Admiral), they are considerably more expensive than corrugated-fin radiators of similar size. This is because instead of just folding the fins back and forth between flat water tubes, round tubes need a soldered fin stack (like most heatpipe air coolers) which is more complicated to manufacture. Reply
  • owan - Wednesday, August 20, 2014 - link

    Well, you're right that those radiators are expensive, likely due to the construction complexity, but they really don't perform anywhere close to accordingly for their cost. I've seen results showing that the Aquacomputer rads perform measurably worse than typical rads at half the price, likely for the reasons C'DaleRider mentioned. Reply
  • garadante - Tuesday, August 19, 2014 - link

    I would really like to see a roundup of the radiators when they're all using the same fan. I was really impressed by the Cooler Master Nepton 280L... Until I noticed fan speed and noise levels. The H110 seems like the best CLC in this roundup by far for having #2/#3 performance at significantly lower fan speeds and noise levels. I know there's the argument for testing CLCs exactly as they come in the packaging, but if you're spending $100+ on a radiator already for a 5-10 C temperature decrease, another $20-30 for a further 5 degree C decrease or even greater isn't a stretch. And if you do such a roundup, having a standard push vs push-pull setup would be great too, to see what the benefits are and which radiators benefit most from the increase static pressure. Reply
  • garadante - Tuesday, August 19, 2014 - link

    And it would be worth looking into if the Corsair H105/similar thick radiators perform better with certain fan setups. Because as it is, the performance is severely lacking in all of the reviews I've seen for the H105 and other thick radiators compared to a 280mm radiator. It would be interesting to find out if the extra radiator thickness comes in handy with enough static pressure or airflow or some combination thereof, or if it's just better to go for more surface area in all use cases. It's much easier to find a case compatible with a 280mm radiator than an extra thick radiator, especially if you want to use push-pull fans. Reply
  • HanzNFranzen - Thursday, August 21, 2014 - link

    I see where you are coming from, but the problem with testing all rads with the same fan is that each radiator performs differently depending on the type of fan used. You could have radiator X that performs better than radiator Y with a Noctua fan and then see the results change in favor of radiator Y when using Corsair fans (just using brands as an example). Combinations of different fan pressures and different radiator thickness and fin densities are all going to skew the results from one fan to another. Reply
  • skrewler2 - Saturday, August 23, 2014 - link

    you are correct, but that's the entire point. also, no one said anything about using only 1 fan. would be nice to see the results with it stock and some of the most popular / performant fans. Then run the tests with the same fans against all the other units. Reply
  • kwrzesien - Wednesday, August 20, 2014 - link

    I'm not sure why he lowered the pump voltage to 7V along with the fans (which seems to be how the article describes it), as far as I can tell most systems leave the pump at a full 12V without any noise issues. Who knows what impact this has on the performance of each of these CLC's but I think it has 0% chance of being used this way in the real world. Reply
  • E.Fyll - Wednesday, August 20, 2014 - link

    The pump voltage is controlled by your motherboard and will be reduced if the thermal control is not disabled in the BIOS, much like it would do with a typical fan. Therefore, I am taking the path that the majority of users would take, which is to leave the BIOS at default settings and allow the thermal control to do its thing.

    Besides, "most" is not "all". For example, the Coolermaster V series is using a pump that is rather noisy. It would not be fair to test some coolers with the pump at maximum speed and some with the pump at a reduced speed, as this obviously has a sizable impact on performance. Everything is being tested under the exact same conditions, otherwise the very use of the word "testing" would be an insult to my ears.
    Reply
  • E.Fyll - Wednesday, August 20, 2014 - link

    Actually, that would be misleading. The performance of the coolers would then rely on the P/V curve of the specific fan used for the test. Each radiator is causing a specific, unique pressure drop and each fan will behave different at each static pressure level. Therefore, the use of a specific fan would favor some designs and cripple others. Switching that fan to another model would yield different results as well. I cannot possibly test every fan out there. I can only be testing each product as a stand-alone solution, it is not possible to account for every possible modification that users may or may not perform. Reply
  • lorribot - Tuesday, August 19, 2014 - link

    You go in to great detail about your test rig and how you test thermal efficiency but completely fail on how you actually test noise, just some " noise measurements are a bit tricky" statement .
    Actually it is not.
    A good test would be the amount of noise produced to cool a given thermal load to say 70C. This would replicate what I would want in the real world, ie how much noise will this thing make cooling my i5/i7 when playing a game or doing massive calculations etc.
    Most noise tests seem arbitrary, such as dB with fans a full speed which is worthless information and provides no useful comparison as all fans run a different speed or are subjective observations of the quality of the noise.
    Noise generated for a number of given workloads would actually be useful please make it happen.
    Reply
  • Hairs_ - Tuesday, August 19, 2014 - link

    Silent pc review provide this sort of analysis, in an anaechoic chamber, with stock vs. reference fans. However, their db classification for "silent" is different, and liquid coolers always perform worse than a good air cooler on a "cooling per dB" metric.

    SPCR's conclusion (which I'd trust above all others) is that liquid coolers are a bad investment as they cost significantly more, produce worse results and are often far too noisy to justify the results. Push enough air and create enough noise, and you can get all sorts of headline grabbing low temperature results...
    Reply
  • Gigaplex - Tuesday, August 19, 2014 - link

    While I agree that air cooling is generally preferable to liquid cooling from an audible efficiency perspective, at the extreme overclocking end of the scale, air coolers just can't keep up. Another downside to the massive tower air coolers is that there's greater risk of damaging the system during frequent transportation (eg LAN party users). There are trade-offs either way you go.

    That said, I prefer air cooling as I optimise my systems for noise.
    Reply
  • AnnihilatorX - Thursday, August 21, 2014 - link

    I disagree. At normal fan speeds liquid coolers tend to lower temperature range and variance, you won't see ultra low temperature, nor you won't see ultra high temperature, and this is due to specific heat capacity of the liquid being higher than air. While overclocked working CPU on air coolers can reach say 70+ deg C, on liquid it would barely reach 40 deg C (speaking from experience) . This is what is important, not low idle temperature. Reply
  • AnnihilatorX - Thursday, August 21, 2014 - link

    To clarify the above reply was replying to Hair not Gigaplex Reply
  • E.Fyll - Wednesday, August 20, 2014 - link

    Actually, it is not as simple as that. Noise measurements are indeed very tricky.

    Your proposed method however is indeed interesting. However, I would need specific equipment to emulate the PWM thermal control of a motherboard and control the coolers in order to maintain a set temperature. I will be adding that in my long "to buy" list.

    I just hope that this will not then start a "comments war" on topics such as "why 150W load and not 160W load", etc etc. :)
    Reply
  • Impulses - Tuesday, August 19, 2014 - link

    Dunno why we're still obsessed with CPU cooling when must enthusiasts are running GPUs that get far hotter and many are running several of them.

    We really need some sorta standard bracket/mount that could facilitate mounting these things to GPUs... Anand has featured a couple but they all seemed to have issues, or maybe I didn't pay close enough attention.
    Reply
  • SantaAna12 - Tuesday, August 19, 2014 - link

    Agree in general about GPUS......but....this is a CPU cooler review that anticipates a new line of enthusiast unlocked chips. Right on time perhaps? Reply
  • abhaxus - Tuesday, August 19, 2014 - link

    I have the Kraken G10 mounting a Kuhler 620 to my reference R9 290. I replaced the Kuhler's fan with the stock fan from an H100 and the radiator is mounted in a rigged up location in my case (have it mounted in the 5.25" bays of my Bitfenix Raider). With the fan on low, it is silent and will only hit 90C if I play an extended session of a high GPU usage game (Crysis 3, BF4). With the fan on medium, which is inaudible with music/game audio playing, I never hit over 70C. This is with the card overclocked to 1107/1350 @ +100mv in Afterburner. My VRM temps stay at around 59-65C as well, and I did NOT put VRM sinks on there, just the standard 92mm fan that comes with the G10.

    Very good purchase overall. Now that the Swiftech H220 is back in America I think I will be purchasing one to properly loop my GPU, however.
    Reply
  • flyingpants1 - Wednesday, August 20, 2014 - link

    I'm totally in agreement, except for the part where you say "many are running several of them". The market for exotic coolers is pretty tiny, and the amount of people who run multiple GPU setups is vanishingly small.

    I'd guess over 90% of PC gamers are using single GPUs under 200 watts, GTX 760 or R9 270x. I'd like to see a slew of water coolers released for those cards.

    The PC market is really, really, reeeeally slow to adapt. It took like 20 years to make USB cables reversible. They're still using ATX, for god's sake. Mini-ITX is nice, but Apple's new Mac Pro slaughters everything from a design standpoint.
    Reply
  • tuxfool - Wednesday, August 20, 2014 - link

    Let me know when you can swap standard parts from a Mac Pro. It sure is nice, but it isn't particularly expandable... Reply
  • kwrzesien - Tuesday, August 19, 2014 - link

    I'm using the H100i myself on the i7-4930K overclocked to 4.5 GHz, so it is probably pushing about 150W under most loads. Cores are hitting 73-78C which I'm fine with and PWM fans are slightly audible but still an enjoyable air-whooshing noise, just enough to know its doing something but not working too hard at it. So I was already wondering if next time around I should look at the 105, 110 or something else - but these charts pretty much say no. The H100i is super-easy to install and fits pretty much any case that tries to handle dual radiators. Even in my Corsair 350D, which could handle the 110, I like having a little room to route cables. I can't see any gains here worth upgrading for, in fact better fans on the 100i would probably match the stock 110 results. Reply
  • jibz - Tuesday, August 19, 2014 - link

    How dare they tarnish the good name of Reserator! I still have my Reserator 2 in a box somewhere (the pump stopped working years ago), and I still like it even tho I stubbed my toes on its metal fins numerous times. Reply
  • Frallan - Wednesday, August 20, 2014 - link

    I agree - the Reserator 1 v2 is a monster I still use it and I acctually plan to purchase one more from a friend that has it unused for my rig (overclocking and crossfire warms the water up to fast to much). But the difference between that and this is enormous - the 1 v2 is passive and completly silent (I struggle to tell if its running with my ear against the tower in a silent room). This brings neither top of the class cooling nor silence - its just a fail.

    just my 0.02€
    Reply
  • AnnihilatorX - Thursday, August 21, 2014 - link

    That's because Reserator 1 has no fan, it is of course compeltely silent :)
    I still have my Reserator 2 XT, while it is nice it is perhaps a bit too bulky for my liking (same applies to Reserator 1). Another advantage of AIO coolers is that it is maintenance free. I have changed the fluid of my Reserator 2 5 times, and each time it is a hassle with the need to go out and buy deionised water. (No distilled water for sale in the UK)
    Reply
  • garadante - Tuesday, August 19, 2014 - link

    Also, I have a question for anyone who can answer. Why is it that according to these tests, most of these CLCs can handle a 340 watt cooling load without breaking 30 C delta T over ambient, yet in the real world you see much higher than this on the temperature of CPUs? According to this, even a massively overclocked 4930k should never break 60 C, but I'm almost certain most users get much higher temperatures than that with these CLCs. Is it because of thermal conductivity between all the intermediate layers between the silicon, paste/solder, IHS, thermal paste, and finally the heatsink/cooling block? Whereas with this testing setup only has the (apparently) copper pad where the heat is initially generated. What causes this significant discrepancy between this synthetic test and real world results?

    I have to wonder just how useful this new synthetic test setup is. Yes, its great for comparing the relative performance of CLCs but it's useless towards showing us real world temperatures we can achieve on actual chips. At least when you have a test bench with a 4930k running at mid 4 GHz, with ambient temperature recorded, you can see what sort of temperatures you can expect to achieve on your own. Because as mentioned above, looking at these results would make me think that if I take a Corsair H110 with the fans running at full speed and put it on my i5-2500k running at 4.5 GHz, the resulting heat load is roughly 100 watts which these results would point towards an extremely low delta T in the high single digit, low double digit range. But if I were to put an H110 on my CPU I would very probably get more around a 30 degree delta T or perhaps even 35 degree. Point being, I like these synthetic tests for comparing relative performance of CLCs in a controlled environment but I would really like a way to relate these results to actual loads. I really like Anandtech and consider your work here generally of very high quality, but I feel it's very disappointing to have to look pretty much anywhere else to find any sort of real world numbers in regard to cooling. I want to know what sort of temperatures I can expect before shelling out $100-120 on a CLC so I don't put that money down and only decrease temperatures 5 C over current temperatures when I'd expect 10-20 C.
    Reply
  • garadante - Tuesday, August 19, 2014 - link

    Correct me if I'm wrong but mounting methodology isn't discussed even once. What thermal paste is used? What curing time is allowed? And in the testing methodology it would lead the reader to believe that the testing results are directly translatable into maximum, worst case real world temperatures of their CPUs. This is absolutely not true as stated before. Take any of these CLCs that have a delta T of 20 C with a 340 watt load and put them on an overclocked -E CPU and your real delta T will be double that, even though the thermal load is significantly below 340 watts. I would love to see this amended and discussed in future reviews because some sort of link to real world performance or at least an acknowledgement that this testing is good for relative performance of CLCs and that's it would be appreciated. Reply
  • garadante - Tuesday, August 19, 2014 - link

    Also, in this testing setup is the heat evenly generated throughout the heatblock? If so that's cause for question, because in CPUs the heat is focused in a comparatively small section towards the center of the IHS, giving the heatsink less surface area to actually transmit heat through. Reply
  • E.Fyll - Wednesday, August 20, 2014 - link

    I am always testing all coolers with the included thermal paste, unless I specifically say otherwise and I would mention the reason why not to. The reason for this is simple; I only care about the products as stand-alone retail items. Trying to fathom every possible usage scenario and modification for each cooler would be an utter waste of time, I would come up with a cure for cancer long before that.

    The test results are perfectly valid and are meant to create a comparative basis for coolers. Yes, they will not tell you the exact temperature of your X CPU, but they will tell you which cooler is and will always be better than another. Besides, you have no idea what the thermal losses of your CPU actually are (no, they are not equal to its energy consumption or with its TDP value, they can be both lower and higher according to the setup, you need a specific device to measure it). Each CPU has an entirely different behaviour depending on over three dozens of factors and even the same exact CPU will report different in-die temperatures due to the nature of the temperature sensing method (it is not exactly perfectly accurate). I cannot possibly test every single CPU out there, let alone many different samples of each.

    The heat is generated from a rod inserted vertically from the base of the copper block, therefore it is concentrated at the center of the block and extending outwards, much like a CPU. The difference is that the sensor which I am using extends across the surface of the block, meaning that it measures the average temperature of the entire surface, not the temperature at a concentrated point at the center of the setup. This is the prime difference between a real CPU and this setup; a real CPU has a very small, pinpointed sensor right where the generation of heat takes place. I could replicate that by placing a very small sensor at the point of the heating rod, but that would ultimately be inaccurate. What the cooler "sees", is the surface it is being mounted on. It does not care about your CPU core, your settings or anything else; it will simply draw energy away from the heatspreader it is in contact with. Whether the CPU itself can transfer the energy to the heatspreader fast enough or anything else, it is case-specific and not related to the actual capability of a cooler. If I were to test different CPUs at the exact same thermal energy output, they would all display different behavior - as a matter of fact, even several pieces of the same exact CPU would do the same exact thing. It goes without saying that I cannot possibly test every possible configuration but, even if I could, that would be meaningless, as the placement of the sensor in the CPU die would be what is controlling the results. I could be testing two CPUs with the exact same thermal energy output, measure the exact same thermal energy being transferred from the heatspreader to the cooler and be reading a difference of dozens of degrees between the two systems, just because they are using a different type of sensor or the sensor is just present at a different location inside the die. This is entirely inaccurate and unscientific, you cannot make any sort of reasonable conclusions based on such results, they would be fabricated. I would rather display the actual capability of each cooler, which is a constant regardless of the CPU it is being mounted on, than provide misleading sets of data.

    So, to summarize, the results depicted in the review are far from useless. They display the actual capability of the cooler, regardless of the CPU or any other external factor. What would be useless is to provide "real data" using a "real CPU", which would be case-specific and misleading for users that do not possess the knowledge to realize that their systems cannot be directly compared.

    Future upgrades may include a device that automatically calculates the °C/W by measuring the exact amount of thermal energy that has been transferred from the block to the cooler, with an integrated sensor at the center of the plate. It is a custom made and rather costly device but it will resemble a CPU even more closely. However, as I am also trying to upgrade our PSU, fan and pretty much all of our testing equipment as well, I cannot build Rome in a day.
    Reply
  • garadante - Thursday, August 21, 2014 - link

    You go through such great lengths to tell my why ever request I suggested or flaw in your testing protocol is wrong. That's just ridiculous. I know you're new to Anandtech, but we're a very diverse crowd here. There are many semi-casual people here who want to become informed regarding certain products as well as many very serious enthusiasts who want to know how various products perform under many conditions. Your testing protocol, quite simply put, only appeals to the very semi-casual readers. And that's being generous. How many consumers will buy a CLC and use it exactly as it comes out of the box? Some, to be sure, but anyone who researches various CPU cooling options will learn that thermal paste is important, and that the bundled thermal paste is quite often sub par. Purchasing a tube of cheap but quality thermal paste (such as the tried and tested Arctic Silver 5) isn't a stretch of the imagination at all, considering it's being paired with $100-160+ CLCs. While swapping fans out is something likely mostly enthusiasts would do, it's still something users would look into if the performance gains are there. Running 2 or 3 tests with each CLC for some different fans (for instance one fan with high static pressure, another with high flow rate, and another with a balance in between) would give invaluable data because it would show users what, if any gains a particular radiator experiences with certain fans. That's not unreasonable at all considering your target audience. But instead you create a 100% synthetic test that while true allows you to produce data such as C/W, is only valid for 100% out of the box CLCs, which seems like it misses the mark. You give data for enthusiasts to calculate the delta C for a particular wattage yet you refuse to admit the validity of and consider implementing an expanded testing protocol to look deeper into the real performance of each radiator (the thing you're really paying money for when you buy a CLC. Fans are cheap)? I've seen your other responses to people suggesting expanded testing procedures or ways to improve the procedure you currently follow, but you always get very defensive with the protocol you've chosen and often times do everything you can to explain in painstaking detail why the reader's suggestion is worthless. This isn't about pride or ego. You're creating content for a large audience that contains many enthusiasts and people very knowledgeable about the subject matter, as well as the semi-casual readers. Yet you produce reviews that are really only useful for the semi-casual reader, completing neglecting the huge fountain of knowledge that is your enthusiast and intellectual reader base. You certainly know a fair amount about how to go about building a reproducible CPU heatsink test bench. But you don't know everything. And many of your readers know quite a bit as well, and some even more than you. Dismissing their suggestions out of hand because of some justification on your part is very narrow minded and undermines the credibility and reputation of this wonderful site as a whole.

    You're new to working for Anandtech, and so you should know that you've joined a well established community. We welcome you gladly to do what you can to raise the standards here even higher, but you cannot expect to do things your way when its against what the community desires. Implementing many of the requests the readers here have made of you should be a trivial thing to strive for as an Anandtech contributor.
    Reply
  • garadante - Thursday, August 21, 2014 - link

    Apologies for the several typos throughout. Curse the lack of an edit button! Reply
  • AnnihilatorX - Thursday, August 21, 2014 - link

    garadante I agree with the E.Fyll here actually. It is far more accurate to compare different coolers using a scientific method, that is, isolate the system and controlled for all other variables. The result in this analysis is valid in that the performance delta between different coolers indicate accurately their relative performance across ALL scenarios, this is because ambient variables such as CPU, systems, are controlled for.

    Yes, the absolute temperature values won't be indicative of a real system, but it doesn't matter, as ambient temperature would have affected different users anyway. No matter how the author sets up the real rig, the results would mean different thing to different people. In this scientific way, it is less work for the author and less variables to control for, and more accurate as well (can't accurately control for CPU power load without special equipment like the one used here).
    Reply
  • garadante - Thursday, August 21, 2014 - link

    But please answer my primary point: testing with uniform fans and uniform thermal paste. Why is this unreasonable to ask for? He's willing to test a dozen and a half CLCs, what I'm saying is there are quite a few people who would be interested in seeing a third of that, just the most interesting ones, with 2 or 3 different types of fans each, focusing on high CFM, high static pressure, and a mixture of the two in order to get a scale of what changes favorably with certain fans for each radiator. Then we can find out if a particular radiator excels at quiet operation with the right fan. Or if another excels at absolute delta T. Or what we can expect to change in performance if we go for a fan a little bit louder but more powerful, or the reverse where it's quieter but a little less powerful. It would show us tradeoffs and illuminate a whole deeper (and extremely relevant) aspect to the world of liquid cooling. Fans with radiators are hugely important but E.Fyll gives it no thought to explore and inform his readers. Reply
  • Death666Angel - Friday, August 22, 2014 - link

    "You're new to working for Anandtech, and so you should know that you've joined a well established community. We welcome you gladly to do what you can to raise the standards here even higher, but you cannot expect to do things your way when its against what the community desires."
    Wow that's condescending. And your assumption to represent the whole of the Anandtech community is quite arrogant as well.
    I personally am fine with his approach and testing methodology. If you want to know the performance of thermal paste, look at one of the tens of reviews that show a difference between the best liquid metal ones and the cheapest of about 5K. And in that narrow a space, differentiating between one thermal paste and another without alternating other factors becomes a problem. I have seen tests where your AS5 performs very badly and others where it is in the middle of the pack. This is due to different application of the paste (distribution and volume) and also the pressure between the heatsink and the IHS. If that varies from one run to another even a little, it can have great consequences when we are talking about fractions of a K determining the position of the paste in the ranking.
    Using a few different fans is a more reasonable request. But then the comments would just shift from "why are you only using the default package stuff" to "why are you not using the fan I am most fond of and instead use these other 3 I don't care for".
    Reply
  • garadante - Saturday, August 23, 2014 - link

    The simple fact is of my many years spent here I've never, ever seen one of the writers at Anandtech post such hostile and condescending posts towards comments of their readers, especially with such regularity as E.Fyll. Perhaps the way I phrased it came off too strong but that's simply what I see and it bothers me, because E.Fyll -is- new here. He was hired on in the last batch, and honestly I'm sad to see him taking over articles from previous writers who did a better, more friendly job reviewing products. Reply
  • DanNeely - Tuesday, August 19, 2014 - link

    You've captured the utter fail of E. Fylladitakis's thermal testing. Absolutely reproducible and absolutely meaningless. Reply
  • garadante - Wednesday, August 20, 2014 - link

    The only usefulness I can see is relative comparison of CLCs. But other than that yeah, no useful meaning from these results as the test stands now. Reply
  • dragosmp - Wednesday, August 20, 2014 - link

    This "only" usefulness is exactly what's needed to decide the best cooler. It may be because I do this as part of my work, but the °/W is the only thing we need to discern between the different cooling configs.

    I have two points to make:
    1. Such a cooler is a complex system and it is described with a number, the °/W. This is simplistic, but sufficient because nobody will add turns to the electric motor of the pump to boost pressure and modify the flow to change the type of flow from laminar to turbulent (if that was even possible). The slope of the rise in °/W with the load as well as the 7V testing (for the pump too if I got it right) points towards the bottleneck. It's unlikely that's the pressure of the pump, and also the type of flow of the air in the radiator fins; I'd point more towards a lack of °/W of the CPU block or too tight tubes.
    2. The temperature delta is of course between the equivalent CPU heat-spreader and the ambient, how could it be any other way? You agree I think that per any CPU architecture the thermal conductivity between the die and the heat-spreader is within very tight margins, from the hat I'd pull a number such as 0.2°/W ==> 20°C of die temp rise for every 100W.

    To sum up: @E.F. - great article, as techie I understand the methodology, the results are perfectly usable. To "translate" these results into what someone might compare to a CPU die temperature it would be enough to put a table with average thermal resistance per architecture of CPU: say SandyBridge 0.15 °/W, Ivy 0.2 °/W, Haswell-E 0.12 °/W etc etc etc and anyone could compute their die temperature from the results you posted. (I give you no one knows precisely what the thermal dissipation actually is in a CPU so prediction of die temp if fraught with peril even when knowing the Rth, but hey...)
    Reply
  • kwrzesien - Wednesday, August 20, 2014 - link

    Now that would be the sort of additional analysis and insight that AnandTech is famous for! In fact not just a table, but a few pages running through test setups on an AMD chip or two (APU & FX) and about five Intel’s: Core i3 Haswell, Core i7 Sandy, Ivy & Haswell (i7-x770K) and a 2011 socket chip like the 4930K or the X (sometimes those seem to be more available to you guys, which is just fantastic for you but the K is the part everyone actually buys - and I'd argue would be the better one to test since it isn't a perfect die so it usually runs at a higher voltage).

    To top it off do each of those at stock and overclocked and then chart all of those TRA's (thermal resistance per architecture - nice job drago!). Then do a i7-4790K just to show that it still sucks.

    I think you would find a few very interesting things. One is that at 32nm the cores have more surface area and thus are able to dissipate their heat better, while the change to 22nm AND the TIM of Haswell has combined to create a localized heat transfer bottleneck between the die and the IHS. There isn't a thing that any air or water cooler can do to help that, you need liquid nitrogen or some kind of freon chilled refrigeration unit that could drop the temperature of the heatsink plate and thus IHS low enough to create a much higher delta T for the die-to-TIM-to-IHS interface to increase the heat transfer rate. Nitrogen isn't practical but I honestly wonder if they couldn't make a Reserator 1 v2-like device that was a mini-AC unit with freon. Of course nobody wants a condensate line running out of their PC, but I'm sure someone could figure something out...

    Or Intel could have made Devils Canyon a product worth buying by soldering the IHS the same way it does on the "E" and Xeon chips. They would have had a big win on their hands if they had done that, even if Haswell still couldn't overclock that well for whatever reason.

    We will know more shortly when Haswell-E drops, which will be 22nm, six or eight Haswell cores and built with a proper soldered IHS. If it's a 22nm heat density issue then E should have a little better heat-transfer properties due to the soldered IHS. If it's a frequency limit with the Haswell cores on 22nm then it will stop at 4.7 GHz (or lower) no matter what. Which will leave the question of whether Broadwell will clock better or have we seen the end of those days?
    Reply
  • bludragon - Wednesday, August 20, 2014 - link

    This is not quite right, since the °/W of the cooler will change depending on the surface area it is attempting to cool. For example, one cooler (lets say with a thin copper base on the water block, but a very large radiator) might work very well across a 30x30mm (900mm2) surface, but if you were to concentrate the heat on a smaller 10x18 (177 mm2) it might work less well than one with a thicker base on the waterblock and a smaller radiator, or even an air cooler which has a thicker base. Reply
  • bludragon - Wednesday, August 20, 2014 - link

    So really what we need is testing using a real cpu since getting heat off the die is the biggest issue with Haswell and I assume the current AMD cpus. It would be good to see that with a stock cpu, and one that has been delidded and had the stock tim replaced with liquidpro. I expect most decent coolers will actually perform within a very similar and narrow range without the delid. I don't know what happens with the delid... Reply
  • garadante - Thursday, August 21, 2014 - link

    Though I beg to pose the question that, if C/W is so important for someone reading these numbers, is it so unreasonable to assume that they may be seriously interested in using fans or thermal paste different from those included in the stock product? Which is one of my primary points in my comments. Testing CLCs 100% stock is useful for a certain segment of the audience but just that, a segment. Those users who care enough about performance to look into a CLC but not those who care enough to do the small extra steps necessary to eke even better performance out of an expensive piece of processor cooling equipment. I argue that the people who care about using better fans and better thermal paste are the same people who care about a radiator's C/W performance. And including a section of the review that looks at the performance of even a subset of these CLCs on actual CPUs would give a valuable contrast for the readers who don't realize the results here on useless in actual applications short of comparing radiators in a controlled environment. A section that says, "Hey, these are the best CLCs from the review but on this and that CPU at stock and such and such overclock the delta T is this. Therefore understand that you cannot use the results here to target a desired temperature for your own setup because there are so many variables. The results of this review simply allow you to choose a CLC based on its merits relative to other CLCs in a high controlled, synthetic environment. You might need research elsewhere if you're trying to hit 60 C Prime95 temperatures on your 4790K running at 4.7 GHz." Reply
  • AnnihilatorX - Thursday, August 21, 2014 - link

    That's actually pretty easy to calculate base on C/W.

    4790k has maximum TDP of 88W, if a cooler has a C/W of 0.07, the cooler would be able to approximately cool the CPU to 6.16 deg C above ambient. You can see the numbers add up by comparing the different thermal watts in the graph "Core Temperature, Constant Thermal Load" and multiplying that with the respective C/W.
    Reply
  • garadante - Thursday, August 21, 2014 - link

    You fail to recognize my points. In the real world any of these CLCs with a C/W of 0.07 or even lower would be lucky to achieve a delta T of 40 with a 4970k at 4.7 GHz. Because there are many more factors involved regarding the temperature of a CPU core. Where the core is located relative to the best contact area between the heatsink and the IHS, what type of TIM is used between the die and the IHS, the IHS and the heatsink. These are things that can only be seen using an actual CPU which is my whole point. It seems that E.Fyll has tricked you into thinking his C/W numbers are at all meaningful in a real application. They serve one purpose and one purpose only: comparing the relative performance of CLCs. But that's easy enough to see just looking at the wattage temperature graphs. No use in making a function out of the results unless you want to know how a CLC would handle under a 1000 watt load on this specific testing rig. Reply
  • garadante - Thursday, August 21, 2014 - link

    And testing with consistent fans across all radiators would give much more useful results when seeing the merits of each radiator. Fans are cheap and it's very reasonable to assume quite a few people would be interested in using aftermarket fans, -if- the performance gains are there, which is my whole point. The Cooler Master Nepton 280L looks good at first when you look solely at the temperature results, but then you look at the fan RPM and dB rating. The Corsair H110 uses a significantly slower and quieter fan but comes within a degree or two delta T of the 280L. Therefore Cooler Master likely uses very noisy, inefficient fans where Corsair uses much more efficient fans and quieter ones at that without sacrificing too much performance. Throw on the same fans on both and you remove that difference and you can see which radiator itself (where your money's going) actually is superior (or at least superior with higher CFM or higher static pressure or a mixture, etc, where E.Fyll coming up with an approximate function would actually be useful). Reply
  • BillyONeal - Wednesday, August 20, 2014 - link

    Great article. It would be nice to see Noctua's NH-D14 or NH-D15 for a comparison with a good air cooler. Reply
  • bludragon - Wednesday, August 20, 2014 - link

    I second that Reply
  • C'DaleRider - Wednesday, August 20, 2014 - link

    Here you go......(hint) It's in the Bench.

    http://www.anandtech.com/bench/CPUCooling/772
    Reply
  • bludragon - Wednesday, August 20, 2014 - link

    Thanks! I had forgotten about that. Only... looking at the results there the temps are not comparable to this article. Is there a post anywhere about the testing methodology for CPU coolers in bench? And yes, I did search... Reply
  • E.Fyll - Wednesday, August 20, 2014 - link

    There will be many reviews of air coolers coming up. As they will be tested using the exact same procedure, the results will be directly compared to those of this review (and every cooling-related review that I perform). Reply
  • mickulty - Wednesday, August 20, 2014 - link

    "Although we cannot fully explain why, it seems that the Reserator 3 Max Dual cannot dissipate high quantities of thermal energy quickly enough."

    Could it be something to do with the overlapping cooling path?
    Reply
  • E.Fyll - Wednesday, August 20, 2014 - link

    Although I cannot really calculate it, I believe that the surface of the pipes is simply much smaller than the surface of a classic two-pass radiator. Therefore, even though it can dissipate heat very quickly via its sizable fin area, the surface between the array and the tubing is not sufficient to maintain a very large energy transfer.

    This simply is an assumption on my part and is not based on any form of evidence.
    Reply

Log in

Don't have an account? Sign up now