Best AIO cooler - is there a consensus?

I was in America over Christmas and managed to get one delivered to the villa we were staying at.

Behold:

H240x01.jpg

If I get time I'll try and install it tonight.

Hopefully it'll get my 5930k comfortably in to the 4.5Ghz range....currently it runs at 4.4Ghz with a Seidon 240m.
 
doyll said:
I like the H220-X .. as well as H140-X and H240-X

The H220-X and H240-X are the only AIO coolers that really match or beat top air coolers. ;)
Click to expand...

Do you think this testing methodology is flawed, i ask becuase i'm deliberating what to buy, and the review suggests that AIOs are better than NH-D14 (i presume is similar to NH-D15)...
 
Kelvin can be added to but pump specs indicate it doesn't have enough power. It stops pumping at 1meter of lift and 72L/H is it's maximum flow rate. That's a whole 17 gallon per hour with no resistance. In other words if you are pumping from a bucket on the ground it won't even push water as high as our chest, and even going into a 5 gallon bucket setting along side the water it would be hard pressed to fill 3 of them in an hour.

No idea how KG is doing it. Part of it is not using similar noise levels for all coolers, but there is more going on. Usually bench testing helps balance things, but obviously there testing didn't. They don't say what the actual temperature of air going into cooler / radiator is. Instead they give the room temperature. Room temperature is rarely the same as cooler / radiator intake and the intake air temp is what we need to use to get accurate comparisons between coolers. Here are results from several tests ran by George Cella. Each time he runs tests, he runs all cooler in the comparison one after the other. So the average I've put in this chart is several different runs at different times for each cooler.

Notice H100i performance and noise level. 73c@68dBA compared to PH-TC14PE 70c@36dBA. Each 10dBA increase doubles how loud it sounds to our ears. 36dBA to 46dBA is twice as loud, 56dBA is 4 time as loud as 36dBA, and 66dBA is 8 times as loud as 36dBA. H100 is 3c warmer than PH-TC14PE and 8 times as loud. Slow the fans down to 40dBA and its 7c warmer. The only AIO that out perform to air are Swiftech H220-X and H240-X .. and then only 5c & 6c respectively, slow down the fans to less than 40dBA and it's 3c & 4c.
CLCvsAirw-H240X_zps9d32853b.jpg


http://www.hitechlegion.com/reviews...loop-280mm-cpu-cooler-review?showall=&start=2
 
Last edited:
Nice one. I sometimes take weeks/months deliberating over parts and layout so this is good to see in the grand scheme of things.
 
doyll said:
Kelvin can be added to but pump specs indicate it doesn't have enough power. It stops pumping at 1meter of lift and 72L/H is it's maximum flow rate. That's a whole 17 gallon per hour with no resistance. In other words if you are pumping from a bucket on the ground it won't even push water as high as our chest, and even going into a 5 gallon bucket setting along side the water it would be hard pressed to fill 3 of them in an hour.

No idea how KG is doing it. Part of it is not using similar noise levels for all coolers, but there is more going on. Usually bench testing helps balance things, but obviously there testing didn't. They don't say what the actual temperature of air going into cooler / radiator is. Instead they give the room temperature. Room temperature is rarely the same as cooler / radiator intake and the intake air temp is what we need to use to get accurate comparisons between coolers. Here are results from several tests ran by George Cella. Each time he runs tests, he runs all cooler in the comparison one after the other. So the average I've put in this chart is several different runs at different times for each cooler.

Notice H100i performance and noise level. 73c@68dBA compared to PH-TC14PE 70c@36dBA. Each 10dBA increase doubles how loud it sounds to our ears. 36dBA to 46dBA is twice as loud, 56dBA is 4 time as loud as 36dBA, and 66dBA is 8 times as loud as 36dBA. H100 is 3c warmer than PH-TC14PE and 8 times as loud. Slow the fans down to 40dBA and its 7c warmer. The only AIO that out perform to air are Swiftech H220-X and H240-X .. and then only 5c & 6c respectively, slow down the fans to less than 40dBA and it's 3c & 4c.
CLCvsAirw-H240X_zps9d32853b.jpg


http://www.hitechlegion.com/reviews...loop-280mm-cpu-cooler-review?showall=&start=2
Click to expand...

thanks for feedback....at some point the greater surface area of a radiator must he able to dissipate more heat at lower noise and directly out of case...so at some point aio coolers will be better...hence i was ready to accept the kg results with a 360mm by 30mm radiator..presumably 2x140 mm radiators should be better than big air and 3x140 would surpass big air.
wonder why 280mm aios are not all better for disipation per dba.
 
ginjaninja said:
thanks for feedback....at some point the greater surface area of a radiator must he able to dissipate more heat at lower noise and directly out of case...so at some point aio coolers will be better...hence i was ready to accept the kg results with a 360mm by 30mm radiator..presumably 2x140 mm radiators should be better than big air and 3x140 would surpass big air.
wonder why 280mm aios are not all better for disipation per dba.
Click to expand...
We are reaching that point with the latest AIO now.
Swiftech H240-X does surpass top air by a slim margin. 1c@same dBA or -4c@+1-6dBA. H220-X is very similar.

There are a couple of CLC knocking at the door, but still seem to have a few shortcomings.
 
ginjaninja said:
Do you think this testing methodology is flawed, i ask becuase i'm deliberating what to buy, and the review suggests that AIOs are better than NH-D14 (i presume is similar to NH-D15)...
Click to expand...

Coolers generally aim to provide an amount of cooling at a noise level, this review totally disregards that. The denser rad on the AIOs give better cooling performance with equal airflow, but creating that airflow makes much more noise (hence the acoustics results so heavily favour air - the H100i test had 100 times as much noise as the NH-D14!! Given background noise will likely be have been close to 30dB an even greater proportion of that noise is likely to be from the cooler than this already substantial difference indicates, but with no background level given that I could see this is hard to verify.) The Noctua could gain performance by using higher RPM fans at the expense of sound, or the AIOs could improve acoustics by dropping their fan speeds. Putting all the fans at 12V is an arbitrary decision (and may actually be harming some of the AIOs quite a bit in terms of acoustics for very little return in thermals).

If you don't care about noise level though and want to use the stock fans at 12V then the testing is sort of OK. However, 10 mins is a short test time, AIOs as with all water-cooling create a time-lag effect in the temperature changes due to the specific heat capacity of water - depending on what they mean about restarting the system this may even be hurting them in the idling test!
 
Last edited:
David Bisset said:
Coolers generally aim to provide an amount of cooling at a noise level, this review totally disregards that. The denser rad on the AIOs give better cooling performance with equal airflow, but creating that airflow makes much more noise (hence the acoustics results so heavily favour air - the H100i test had 100 times as much noise as the NH-D14!! Given background noise will likely be have been close to 30dB an even greater proportion of that noise is likely to be from the cooler than this already substantial difference indicates.) The Noctua could gain performance by using higher RPM fans at the expense of sound, or the AIOs could improve acoustics by dropping their fan speeds. Putting all the fans at 12V is an arbitrary decision (and may actually be harming some of the AIOs quite a bit in terms of acoustics for very little return in thermals).

If you don't care about noise level though and want to use the stock fans at 12V then the testing is sort of OK. However, 10 mins is a short test time, AIOs as with all water-cooling create a time-lag effect in the temperature changes due to the specific heat capacity of water - depending on what they mean about restarting the system this may even be hurting them in the idling test!
Click to expand...

my reading of the review say at a given cooling performance (where s36 outperforms nh-d14), the Dba is the same 30/30.2, albeit this 10 min factor might be skkewing it..but as it stands the review is showing better performance per dba for s36. i do care about noise and 30Dba is more than enough...so from this feedback i will be sticking with NH-D15.
 
Dave, how do you figure 100 times as loud? I was under the impression that 70dBA sounds 8 times as loud as 40dBA because our ears perceived a 10dBA increase to sound twice as loud, and as each 10dBA doubles the sound to our ears 70db is 8 times as loud as 40dBA .. or 16 times as loud as 30dBA.

Indeed, the more coolant in the system the longer it take it to reach operating temperature when heating up and cooling off. Air systems can bring CPU to maximum or minimum in a couple of minutes, but H2O can take 20-30 minutes. Any time an air cooler takes more than 5 minutes to hit peak or minimum the airflow into cooler is heating up / cooling down along with it. This is because the heated airflow from cooler is circling around into cooler's cool intake air and raising it's temperature.
 
ginjaninja said:
my reading of the review say at a given cooling performance (where s36 outperforms nh-d14), the Dba is the same 30/30.2, albeit this 10 min factor might be skkewing it..but as it stands the review is showing better performance per dba for s36.
Click to expand...

The s36 does indeed look better, I'd not argue that :) I used the H100i as a data point to highlight why the testing methodology was broken. As airflow is very complex you may get both better cooling and better acoustics using a different fan voltage as smoother flow can have gains for both. From those results, it looks like the s36 is a very good choice - just wanted to highlight the problems with the method.

doyll said:
Dave, how do you figure 100 times as loud? I was under the impression that 70dBA sounds 8 times as loud as 40dBA because our ears perceived a 10dBA increase to sound twice as loud, and as each 10dBA doubles the sound to our ears 70db is 8 times as loud as 40dBA .. or 16 times as loud as 30dBA.
Click to expand...
The decibel scale is logarithmic - so a noise that is 10dB louder is 10 times as powerful, 20dB higher is 100 times as powerful. Our ears are weird so perceived increases may vary depending on lots of factors like frequency etc, and I've no idea how these are taken account of in the testing. A common interpretation is that 10 times as much power is felt as twice as loud to the human ear, so you could conclude it's 4 times as loud - I should have said it's generating 100 times as much noise energy really I guess? Words :(

doyll said:
Indeed, the more coolant in the system the longer it take it to reach operating temperature when heating up and cooling off. Air systems can bring CPU to maximum or minimum in a couple of minutes, but H2O can take 20-30 minutes. Any time an air cooler takes more than 5 minutes to hit peak or minimum the airflow into cooler is heating up / cooling down along with it. This is because the heated airflow from cooler is circling around into cooler's cool intake air and raising it's temperature
Click to expand...
Yep - easy to forget that air can keep going up too! Good point. Another case where the testing isn't great, though it's hard to say who that's favouring! The bench is presumably open (I think it said but internet here is bad so not going to reopen it) which will be giving the air coolers a boost after the first few mins by keeping plenty of cool air fed, real cases may be too limited in airflow.
 
Last edited:
I was just re-reading the KG testing procedure. They talk in circles.
To test the performance of the Fractal Design Kelvin S36, we built a powerful X79 based system on an open test bench to push the cooler to the limits. As we were testing on an open test bench, we directed a Be Quiet! Shadow Wings 120 mm fan at the VRM circuitry to better simulate the airflow generated in a case.
Later they say:
In our noise level tests we switched off all the case fans and replaced the AMD Radeon HD 7950 graphics card with a HD 5550 passive model so were just taking into account the noise generated by the cooler fans themselves.
Their open bench testing has have case fans? ! ? :eek:

30dBA is extremely quiet I don't have a room in my house, even in the basement with nothing at all to make noise, that is below 30dBA on a meter.

Therefore I question their sound levels. NH-D14 is usually 38-40dBA at full speed. Huge difference to their sound levels.

H100i in comparison tests with top air with same temp of air going into cooler/radiator give similar temps as top air, yet KG show it 4.8-4.8c better cooling.

I know George Cella and know how he tests. His results are always consistent and dependable .. like others testers who are very careful to test with lots of attention to the veriables and who keep all possible variables as close to the same as possible and get similar results.
 
Last edited:
David Bisset said:
The decibel scale is logarithmic - so a noise that is 10dB louder is 10 times as powerful, 20dB higher is 100 times as powerful. Our ears are weird so perceived increases may vary depending on lots of factors like frequency etc, and I've no idea how these are taken account of in the testing. A common interpretation is that 10 times as much power is felt as twice as loud to the human ear, so you could conclude it's 4 times as loud - I should have said it's generating 100 times as much noise energy really I guess? Words :(
Click to expand...
Just as I figured, but wanted to be sure. :D

David Bisset said:
Yep - easy to forget that air can keep going up too! Good point. Another case where the testing isn't great, though it's hard to say who that's favouring! The bench is presumably open (I think it said but internet here is bad so not going to reopen it) which will be giving the air coolers a boost after the first few mins by keeping plenty of cool air fed, real cases may be too limited in airflow.
Click to expand...
I hate tester who do not monitor the cooler / radiator intake air temp .. even worse when they use a case too. Testing using room ambient instead of actual intake air temp leave testing open to a huge variable in the actual air temp going into cooling component. Temps around a bench with all the heat coming off of system components as well as one or two people radiating heat is not the same as the thermostat on the opposite wall.

It's not complicated or expensive to monitor the intake air temp,
but because
"it's not the way most reviewers do it" or
"we test in a case to give real world comparisons" (only real world is theirs as our systems are different) or
"cooling intake air temp changes too much to be used" (yeah, I've been told that too)​
they use room ambient.

Put the cooler in a case and cooler intake air temp can easily be 10-15c warmer 10 minutes into a run. Even more if cooler has high performance fans like Thermalright "Extreme" coolers have.
 
Last edited:
^ Pretty much. A cooler review should measure the cooler fans intake temp. Getting the intake temp low is the responsibility of the user, the case, the case fans and the room ambient- shouldn't really be a part of a cooler review other than a warning about what would happen. AIOs are less susceptible to case temps usually since they can be placed in areas of fresh intake on a case by case basis- especially so with custom loops.
 
I'm so confused with the testing results coming out on the Kraken X61 and the H 240-x some reviews have the 240x as being middle of the pack noisy for CLCs but then others have it quieter than the DH 15 and be quiet dark rock pro 3 air coolers which are apprenretly quieter than a nun's farts.

If I want to build a big tower (p380 or primo) build and have gone to all the effort of ripping case fans out and replacing them with silent ones the least the ******* reviewers could do is give a better noise level reading.

http://www.silentpcreview.com/NZXT_Kraken_X61


The silent PC review has the x61 on par with the dh15 with slightly more noise which I can live with in exchange for actually being able to mount ram on a motherboard (****ing about with huge tower coolers and ram is the main reason i'm consdering CLCs) and they really are not a fan of the H240 at the price it is

http://www.silentpcreview.com/article1428-page1.html

But then even they admit they've not done the reviews in a f'ing case.

I'm utterly bamboozeled, i'm not so fussed about the best performance, i'm happy with DH12/15 levels without the faffing witht he tower - but none of the reviews make it clear if the AIOC are actually as quiet.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom