Fyll, it's possible that you've covered this before, but in cases where AIO coolers have the same heatsink/block unit as these two do, is the absolute thermal resistance essentially measuring the effectiveness of their radiators and fans, or are there other potential variables to consider?
I'm guessing, for instance, that the days of significant variations in the quality of machining of the heatsink contact surface are long over? If I never have to lap another heatsink, it'll be too soon...
Thermal resistance is affected by: 1) Overall sink and radiator design 2) Fluid flow rate 3) Fluid type (not much change since all use basically the same thing) 4) Fan speed and efficiency 5) Mounting pressure 6) Thermal paste quality
Assuming 1-3 are identical (because of OEM designs), it's fan and paste performance and mounting pressure you see change thermal conductance
Nothing in this review shows a need to replace my Noctua NH-D15, cooling an i7 5960X @ 4.3 and an Asus STRIX GTX 980Ti. I'm having absolutely no problem with heat or air flow. One of these days I'll take the plunge and try a water AIO, but this ain't the day.
So if "I'm having absolutely no problem with heat or air flow" then why did you read this whole thing for? ;) I much rather have an AIO for a clean look, easier to clean than a big air cooler and trust me...I used to love them! I had the D14 with 3 massive fans AND shrouds before. I started the thread for it in case you are curious how it used to look: http://www.overclock.net/t/628569/official-noctua-...
Not saying that Air coolers are not good...the D15 is the top in my list but I prefer AIO.
A good air cooler is still better than most of the entry level/mid segment AIOs. AIOs have a copper heat plate and aluminium radiator, couple that with non-refillable system you might face a failed pump due to corroded metal. If you want an AIO better be ready to spend on likes of Swiftech/EKWB one as they don't mix metals together in the system and are user serviceable and expandable.
Great cooling, but it needs to get at least 10dB quieter to be comfortably usable, 48dB is way too much to be comfortable for me. Can't claim I have any idea how to reduce it that much though (except having luck with better pump than the one with whinning noise as mentioned in review)...
/wishful thinking: Tbh, house air condition systems imho could already provide bayonet plugs for PC water cooling, let the noise and heat go where it doesn't hurt :-)
An AIO cooler that's too loud, too expensive, too bulky, and under-performs to much cheaper, much quieter, much smaller traditional HSF? Color me shocked. Stupid bling for stupid people.
These AIO coolers only under-perform traditional HSFs that are themselves quite expensive and bulky, so the situation is not quite as black-and-white as you suggest.
I agree with the other commenters: can we see what the overall noise/performance is like compared to very high-end air coolers, like the NH-D15, the NH-U14S, the PH-TC14PE, etc.?
I would love it if AnandTech would get their hands on both of the EK Predator models. I've heard their pump noise is a lot better, not to mention expandable (the 360) and refillable.
I bought the H100i GTX for my last build a couple months ago in a Define R5 Chassis (Has sound dampening). I cannot hear any part of the system outside the case and I can barely hear it if I stick my head inside the case on balanced mode. Performance mode you can of course hear it even outside the case. But then again that also puts it at 100%. And I can pretty much guarantee that with a mid to high end graphics card, you won't be able to hear this over whatever noise the graphics subsystem is putting out. The only part I ever hear is my GTX 970 spinning up.
I recently used the H80i GT in a new Skylake i7 6700k build and I am very happy with it. Very happy! It is quiet as a mouse and was relatively easy to install. I have a large Corsair Air 540 case, so the size of the unit was not an issue for me. I'll have to say that I am still old school and don't trust that factory TIM they pre-apply. I always remove it with alcohol and apply a pea sized dollop of MX-4 on the CPU die before mounting the water block. I am amazed everyday when I use this rig that it idles so cool. Right now RealTemp shows 14/13/12/15 (C). Absolutely great cooler. One thing you didn't realty talk too much about was the Corsair Link software. This software caused a slew of issues for me, including crashes, lockups/freezes, and a dozen errors in my Event Viewer. I uninstalled and reinstalled the software until I finally got fed up with it and completely uninstalled it and detached the USB cable from the pump. The cooler works great as I already stated, and I can still control the fans in my BIOS so no biggie I guess, but this Corsair Link leaves something to be desired. It's buggy and not worth the headache to have a multi-colored logo on the pump.
What was your ambient temperature (give or take) while you measured those temperatures? Because 12-15°C sounds more like a peltier+compressor cooler...
Oops soory cut off answer there. It has to be a combination of this cooler working great, an extremely cool running processor, and my basement office staying cool 24/7. It's about 75F outside right now and my office is still around 20-21C and my idle temps are 15/13/12/13 C. It is AWESOME!
I rather suspect there is something wrong with your measurement tool (does CoreTemp64 show same values?), otherwise you are 6-9 degrees below ambient (which shouldn't be possible without forced cooling). Such a temperature difference quite rings water condensation warning bells, depending on humidity and air flow around cpu block.
Yeah sorry to burst your bubble but there must be something wrong with the measurement as HollyDol suggested. You cannot have lower temps than ambient with an AIO!
Ok - so I rechecked with AIDA 64 Sensor Panel and it shows temps a little higher, but still very good. Real good. Still below 20C. The digital thermometer on my wall next to a window reads 20C.
okay, that's officially weird. I'd be tempted to put thermometer on cpu block to get alternate readings. And, if it is still that low, sell that cpu back to Intel for lots of $$$.
Oh, I never doubted your honesty about the numbers you're getting from realtemp, or aida, or whatever. It doesn't matter what they tell you though; if you're getting below ambient on an AIO or air cooler... it's wrong, because the medium you are cooling with is the air in that room.
Good review, but I'd really like to see two things. First if you could pick a good noise level for use without headphones and normalize to that if possible for the product so we could see what lowering some of the more aggressively tuned coolers to a level that's bearable to listen to. Second, I'd love to see whether or not the coolers can be set to an inaudible or at least very quiet idle setting, and if so how much heat they can dissipate effectively in that setting.
Adding those and mixing in some air coolers (probably one or two of the high end kilo of metal hanging off your socket coolers, and a stock cooler and 212 as a baseline) would make the reviews fantastic and a source for all the information I'd want when looking at coolers.
+1 for the first request... pick a nice acceptable target level in dBA and a fairly average OC'd CPU thermal load and see what temps the coolers can manage. The chart topping temps for these coolers are completely irrelevant for the overwhelming majority of users because of the noise levels. I suspect most coolers would produce impressive temps if they have a jet engine pushing air through them.
I realise it's not a perfect comparison, dBA is not a perfect measure of how intrusive the noise levels are, and no doubt people would quibble over whatever dBA and wattage you decided on. BUT, sometimes you just want a quick chart that gives a reasonably accurate reflection of performance under relatively "normal" enthusiast workloads, and temps over ambient at a fixed dBA & wattage would give you that.
To the author - I really appreciate your scientific approach. Thanks for the article and I look forward to seeing the comparison with the high Noctua's in future too.
Is 1400RPM the lowest these coolers will get using the Corsair Link software? My H100i idles at ~900RPM using a Swiftech PWM hub and is not only inaudible from ~2-2.5 feet away, but also maintains sub-30*C idle temps in a 23*C room with a 4970k. Even at load, the cooler is inaudible (or at least not noticeable) while wearing a pair of open-back headphones.
If these new coolers are indeed quieter than the preceding H100i/H80i, then "quiet but tolerable" seems like it is selling them short. You can make these things silent with a custom fan curve and still achieve adequate cooling.
As a side note, I do think the older H100i/H80i have a much more sleek and attractive appearance. The two-tone grey/black on these new ones looks plasticky and cheap.
Such a pity the H110i GTX wasn't in the mix, while you were at it. That really seems like the one worth looking at for anyone who has room for it. Let's face it, if you are in the market for this much cooling, you're probably in the market for as much as you can get for the least noise, which is the H110i GTX.
Still, after seeing the figures and description of the H100i GTX pump noise I'm more comfortable sticking with big air for my next build.
There are a couple of real world tests that seems to be missing but may actually be quite useful, firstly cooling performance at a given noise level say 35/40/45db rather than at a given voltage or or how noisy is a system to maintain a 200w load at a say a temperature of 80C.
www.silentpcreview.com constant ambient conditions during their testing are 10~11 dBA and 21~23°C. Anything above 20 dBA is rightfully considered loud.
You need to learn your lesson and stop bullshitting with the following laughable arguments, cause you clearly don't know what inaudible and loud really means: - "The noise floor of our recording equipment is 30.2-30.4 dB(A), which represents a medium-sized room without any active noise sources."; - "<35dB(A) = Virtually inaudible"; - "35-38dB(A) = Very quiet"; - "38-40dB(A) = Quiet".
Is undervolting the fans the right approach for pwm based fans? i would have thought that using the control software would be the way to control the rpm for better noise control? This may not work for your standardized tests though.
I am a Corsair fanboy, I will admit it with. Every aftermarket components from fans to case to keyboard is Corsair. However, I have never been impressed with their AIO coolers. Every model I have tried over the years have all resulted in NOISY pumps after a few days. In the case of the H115 it literally took one stress test under load for the pump to start clicking.
I have been using my Zalman CNPS20LQ for years with my i7 3770K (push/pull with 2x Corsair SP120) and it's performance has remained constant since day one. Pump is whisper quiet, and my own tests put it between the H80 and H100i (which I have purchased to replace the Zalman but ended up returning).
Personally I switched to watercool due to noise levels back in C2D + 8800GT days... having one pump sitting in a "foam" to cancel vibrations and big passive radiator outside the chasis = computer silent enough to have it crunching numbers over night and sleep just next to it. And I am still completely happy with the solution. Today it cools my i5-2500K and GTX580 and most likely will be cooling next machine when time comes and money allows...
Not that you would have known but the Corsair Link software is a mess. They've done a lot to clean it up but it still has major issues. You can check their forums if you don't believe me. I believe it's also why you can't find their LED lighting kits or commanders anymore. I'm happy with my H100i but was pretty disappointed with the software.
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
47 Comments
Back to Article
zeeBomb - Monday, November 16, 2015 - link
I came for the coolers. I wanna stay COOL.Der2 - Monday, November 16, 2015 - link
A cooling system is essential for a great PC build.ShortWicky - Monday, November 16, 2015 - link
Brilliant deduction Einstein, do go on with more glimmers of genius insightBlack Obsidian - Monday, November 16, 2015 - link
Fyll, it's possible that you've covered this before, but in cases where AIO coolers have the same heatsink/block unit as these two do, is the absolute thermal resistance essentially measuring the effectiveness of their radiators and fans, or are there other potential variables to consider?I'm guessing, for instance, that the days of significant variations in the quality of machining of the heatsink contact surface are long over? If I never have to lap another heatsink, it'll be too soon...
basroil - Monday, November 16, 2015 - link
Thermal resistance is affected by:1) Overall sink and radiator design
2) Fluid flow rate
3) Fluid type (not much change since all use basically the same thing)
4) Fan speed and efficiency
5) Mounting pressure
6) Thermal paste quality
Assuming 1-3 are identical (because of OEM designs), it's fan and paste performance and mounting pressure you see change thermal conductance
nathanddrews - Monday, November 16, 2015 - link
What AIO coolers have the quietest pumps (least whine)?rpjkw11 - Monday, November 16, 2015 - link
Nothing in this review shows a need to replace my Noctua NH-D15, cooling an i7 5960X @ 4.3 and an Asus STRIX GTX 980Ti. I'm having absolutely no problem with heat or air flow. One of these days I'll take the plunge and try a water AIO, but this ain't the day.Thanks for the great review!
shaolin95 - Monday, November 16, 2015 - link
So if "I'm having absolutely no problem with heat or air flow" then why did you read this whole thing for? ;)I much rather have an AIO for a clean look, easier to clean than a big air cooler and trust me...I used to love them! I had the D14 with 3 massive fans AND shrouds before. I started the thread for it in case you are curious how it used to look:
http://www.overclock.net/t/628569/official-noctua-...
Not saying that Air coolers are not good...the D15 is the top in my list but I prefer AIO.
Chaitanya - Monday, November 16, 2015 - link
A good air cooler is still better than most of the entry level/mid segment AIOs. AIOs have a copper heat plate and aluminium radiator, couple that with non-refillable system you might face a failed pump due to corroded metal. If you want an AIO better be ready to spend on likes of Swiftech/EKWB one as they don't mix metals together in the system and are user serviceable and expandable.Morawka - Tuesday, November 17, 2015 - link
AIO are disposable.. you use then for 3-5 years then that's it, 50% water has evaporated. You could open it up, but good luck with that.HollyDOL - Monday, November 16, 2015 - link
Great cooling, but it needs to get at least 10dB quieter to be comfortably usable, 48dB is way too much to be comfortable for me. Can't claim I have any idea how to reduce it that much though (except having luck with better pump than the one with whinning noise as mentioned in review).../wishful thinking:
Tbh, house air condition systems imho could already provide bayonet plugs for PC water cooling, let the noise and heat go where it doesn't hurt :-)
GreenMeters - Monday, November 16, 2015 - link
An AIO cooler that's too loud, too expensive, too bulky, and under-performs to much cheaper, much quieter, much smaller traditional HSF? Color me shocked. Stupid bling for stupid people.Black Obsidian - Monday, November 16, 2015 - link
These AIO coolers only under-perform traditional HSFs that are themselves quite expensive and bulky, so the situation is not quite as black-and-white as you suggest.jabber - Monday, November 16, 2015 - link
Still got a H50 in the cupboard. Having switched to non-homebrew PCs recently its been waiting for a new project.ikjadoon - Monday, November 16, 2015 - link
I agree with the other commenters: can we see what the overall noise/performance is like compared to very high-end air coolers, like the NH-D15, the NH-U14S, the PH-TC14PE, etc.?MrTeal - Monday, November 16, 2015 - link
If you can, it would be very interesting to see how the EK Predator or Swiftech H240x does on your testbench relative to the AOI coolers.thestryker - Monday, November 16, 2015 - link
I would love it if AnandTech would get their hands on both of the EK Predator models. I've heard their pump noise is a lot better, not to mention expandable (the 360) and refillable.Pissedoffyouth - Monday, November 16, 2015 - link
I'm thinking I might use one of these with one of those Broadwell APU's in a really tiny case. Should be able to overclock the iGPU massivelymaximumGPU - Monday, November 16, 2015 - link
And that is why i switched back to air cooling, couldn't stand that pump noise!Eagle1848 - Monday, November 16, 2015 - link
I bought the H100i GTX for my last build a couple months ago in a Define R5 Chassis (Has sound dampening). I cannot hear any part of the system outside the case and I can barely hear it if I stick my head inside the case on balanced mode. Performance mode you can of course hear it even outside the case. But then again that also puts it at 100%. And I can pretty much guarantee that with a mid to high end graphics card, you won't be able to hear this over whatever noise the graphics subsystem is putting out. The only part I ever hear is my GTX 970 spinning up.hapkiman - Monday, November 16, 2015 - link
I recently used the H80i GT in a new Skylake i7 6700k build and I am very happy with it. Very happy! It is quiet as a mouse and was relatively easy to install. I have a large Corsair Air 540 case, so the size of the unit was not an issue for me. I'll have to say that I am still old school and don't trust that factory TIM they pre-apply. I always remove it with alcohol and apply a pea sized dollop of MX-4 on the CPU die before mounting the water block. I am amazed everyday when I use this rig that it idles so cool. Right now RealTemp shows 14/13/12/15 (C). Absolutely great cooler. One thing you didn't realty talk too much about was the Corsair Link software. This software caused a slew of issues for me, including crashes, lockups/freezes, and a dozen errors in my Event Viewer. I uninstalled and reinstalled the software until I finally got fed up with it and completely uninstalled it and detached the USB cable from the pump. The cooler works great as I already stated, and I can still control the fans in my BIOS so no biggie I guess, but this Corsair Link leaves something to be desired. It's buggy and not worth the headache to have a multi-colored logo on the pump.HollyDOL - Monday, November 16, 2015 - link
What was your ambient temperature (give or take) while you measured those temperatures? Because 12-15°C sounds more like a peltier+compressor cooler...hapkiman - Monday, November 16, 2015 - link
Ambient in my office is about 21C or around 70F. I know - my best friend didn't believe me either and he came aND SAW.hapkiman - Monday, November 16, 2015 - link
Oops soory cut off answer there. It has to be a combination of this cooler working great, an extremely cool running processor, and my basement office staying cool 24/7. It's about 75F outside right now and my office is still around 20-21C and my idle temps are 15/13/12/13 C. It is AWESOME!HollyDOL - Tuesday, November 17, 2015 - link
I rather suspect there is something wrong with your measurement tool (does CoreTemp64 show same values?), otherwise you are 6-9 degrees below ambient (which shouldn't be possible without forced cooling). Such a temperature difference quite rings water condensation warning bells, depending on humidity and air flow around cpu block.maximumGPU - Tuesday, November 17, 2015 - link
Yeah sorry to burst your bubble but there must be something wrong with the measurement as HollyDol suggested. You cannot have lower temps than ambient with an AIO!Beararam - Tuesday, November 17, 2015 - link
Real temp is drunk. At the risk of being redundant, no AIO is going to get you below ambient. Not possible.hapkiman - Tuesday, November 17, 2015 - link
Ok - so I rechecked with AIDA 64 Sensor Panel and it shows temps a little higher, but still very good. Real good. Still below 20C. The digital thermometer on my wall next to a window reads 20C.17/18/15/16 on Aida 64 Extreme sensor panel.
hapkiman - Tuesday, November 17, 2015 - link
http://i1116.photobucket.com/albums/k561/hapkiman/...HollyDOL - Wednesday, November 18, 2015 - link
okay, that's officially weird. I'd be tempted to put thermometer on cpu block to get alternate readings. And, if it is still that low, sell that cpu back to Intel for lots of $$$.Beararam - Wednesday, November 18, 2015 - link
Oh, I never doubted your honesty about the numbers you're getting from realtemp, or aida, or whatever. It doesn't matter what they tell you though; if you're getting below ambient on an AIO or air cooler... it's wrong, because the medium you are cooling with is the air in that room.weewoo87 - Thursday, September 15, 2016 - link
Is your PC on the floor next to your AC vent? The air coming out of your AC vent will be cooler than ambient.xthetenth - Monday, November 16, 2015 - link
Good review, but I'd really like to see two things. First if you could pick a good noise level for use without headphones and normalize to that if possible for the product so we could see what lowering some of the more aggressively tuned coolers to a level that's bearable to listen to. Second, I'd love to see whether or not the coolers can be set to an inaudible or at least very quiet idle setting, and if so how much heat they can dissipate effectively in that setting.Adding those and mixing in some air coolers (probably one or two of the high end kilo of metal hanging off your socket coolers, and a stock cooler and 212 as a baseline) would make the reviews fantastic and a source for all the information I'd want when looking at coolers.
rhysiam - Monday, November 16, 2015 - link
+1 for the first request... pick a nice acceptable target level in dBA and a fairly average OC'd CPU thermal load and see what temps the coolers can manage. The chart topping temps for these coolers are completely irrelevant for the overwhelming majority of users because of the noise levels. I suspect most coolers would produce impressive temps if they have a jet engine pushing air through them.I realise it's not a perfect comparison, dBA is not a perfect measure of how intrusive the noise levels are, and no doubt people would quibble over whatever dBA and wattage you decided on. BUT, sometimes you just want a quick chart that gives a reasonably accurate reflection of performance under relatively "normal" enthusiast workloads, and temps over ambient at a fixed dBA & wattage would give you that.
To the author - I really appreciate your scientific approach. Thanks for the article and I look forward to seeing the comparison with the high Noctua's in future too.
thekdub - Monday, November 16, 2015 - link
Is 1400RPM the lowest these coolers will get using the Corsair Link software? My H100i idles at ~900RPM using a Swiftech PWM hub and is not only inaudible from ~2-2.5 feet away, but also maintains sub-30*C idle temps in a 23*C room with a 4970k. Even at load, the cooler is inaudible (or at least not noticeable) while wearing a pair of open-back headphones.If these new coolers are indeed quieter than the preceding H100i/H80i, then "quiet but tolerable" seems like it is selling them short. You can make these things silent with a custom fan curve and still achieve adequate cooling.
As a side note, I do think the older H100i/H80i have a much more sleek and attractive appearance. The two-tone grey/black on these new ones looks plasticky and cheap.
nmm - Monday, November 16, 2015 - link
Such a pity the H110i GTX wasn't in the mix, while you were at it. That really seems like the one worth looking at for anyone who has room for it. Let's face it, if you are in the market for this much cooling, you're probably in the market for as much as you can get for the least noise, which is the H110i GTX.Still, after seeing the figures and description of the H100i GTX pump noise I'm more comfortable sticking with big air for my next build.
Awful - Monday, November 16, 2015 - link
Still using my original H100 on i2500K @ 4.8GHz. Still see no good reason to upgrade either. Nice!lorribot - Monday, November 16, 2015 - link
There are a couple of real world tests that seems to be missing but may actually be quite useful, firstly cooling performance at a given noise level say 35/40/45db rather than at a given voltage or or how noisy is a system to maintain a 200w load at a say a temperature of 80C.mobutu - Tuesday, November 17, 2015 - link
www.silentpcreview.com constant ambient conditions during their testing are 10~11 dBA and 21~23°C.Anything above 20 dBA is rightfully considered loud.
You need to learn your lesson and stop bullshitting with the following laughable arguments, cause you clearly don't know what inaudible and loud really means:
- "The noise floor of our recording equipment is 30.2-30.4 dB(A), which represents a medium-sized room without any active noise sources.";
- "<35dB(A) = Virtually inaudible";
- "35-38dB(A) = Very quiet";
- "38-40dB(A) = Quiet".
HollyDOL - Tuesday, November 17, 2015 - link
Please review your lessons of physics or at least use google before you start accusations, one of many: http://www.cdc.gov/ncbddd/hearingloss/sound.htmlBeararam - Tuesday, November 17, 2015 - link
"cause you clearly don't know what inaudible and loud really means""Anything above 20 dBA is rightfully considered loud."
The whispers! Stop the whispers, my ears are bleeding! Oh the pain!
Maybe you need to walk around with earmuffs on.
samsp99 - Wednesday, November 18, 2015 - link
Is undervolting the fans the right approach for pwm based fans? i would have thought that using the control software would be the way to control the rpm for better noise control?This may not work for your standardized tests though.
Bansaku - Wednesday, November 18, 2015 - link
I am a Corsair fanboy, I will admit it with. Every aftermarket components from fans to case to keyboard is Corsair. However, I have never been impressed with their AIO coolers. Every model I have tried over the years have all resulted in NOISY pumps after a few days. In the case of the H115 it literally took one stress test under load for the pump to start clicking.I have been using my Zalman CNPS20LQ for years with my i7 3770K (push/pull with 2x Corsair SP120) and it's performance has remained constant since day one. Pump is whisper quiet, and my own tests put it between the H80 and H100i (which I have purchased to replace the Zalman but ended up returning).
MamiyaOtaru - Thursday, November 19, 2015 - link
CPU watercoolers seem so anachronistic to me (like BTX mobos) now that GPUs are using far more powerHollyDOL - Saturday, November 21, 2015 - link
Personally I switched to watercool due to noise levels back in C2D + 8800GT days... having one pump sitting in a "foam" to cancel vibrations and big passive radiator outside the chasis = computer silent enough to have it crunching numbers over night and sleep just next to it. And I am still completely happy with the solution. Today it cools my i5-2500K and GTX580 and most likely will be cooling next machine when time comes and money allows...6SpeedRobbyG - Saturday, November 21, 2015 - link
Can anyone confirm the H80i GT will fit in my Cooler Master Elite 130?hkuspc40 - Sunday, November 22, 2015 - link
Not that you would have known but the Corsair Link software is a mess. They've done a lot to clean it up but it still has major issues. You can check their forums if you don't believe me. I believe it's also why you can't find their LED lighting kits or commanders anymore. I'm happy with my H100i but was pretty disappointed with the software.