• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

TRUE vs WATER: Comparison on OC'd i7 Platform.

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

baditude_df

Northern Senior
Joined
Apr 24, 2007
As promised, here my comparison of a High-End Water Cooling loop to the Thermalright Ultra 120 Exteme (TRUE) air cooled heatsink.
This comparison is strictly to test the the noise and thermal performance of the TRUE against that of a quality water cooling loop in an environment that is as controlled as possible without having laboratory conditions or equipment.
However, practicality's sake, a line will be drawn on both data sets when the threshold of acceptable noise level is reached and the comparison passes into the realm of straight up thermal efficiency. Noise results will be qualitative and not quantitative, as I have no means of directly measuring noise.
I will not be testing any other loop or configuration than the one I already own and have set up. Have a look below and you will see that this is a properly configured loop comprised of high end components. The machine sits 3 feet from my head when seated at my desk.

Test Setups:

General:

  • Both air and water will be in open air. When reference is made to Ambient Temperature, that is actually Ambient ROOM Temperature.
  • System Specs (Frequeny, Votage, HT etc.) are in my sig. Cooling Specs below.
  • If you want a very rough breakdown of Fan Specs at each Voltage setting just divide into 4 equal divisions.
    ie: 12v=2600RPM,102cfm,39dBa,64Pa 10v=2000RPM,75cfm,29dba,48Pa 8v=1500RPM, 50cfm,19dba,36Pa 6v=1000RPM,25cfm,10dBa,20Pa
    Obviously fans don't scale this perfectly, but from what I could feel and hear during this comparison, it shouldn't be too far off.

AIR:

imgp0560d.jpg


HS: 1 TRUE Black - Original (No Revision)
FAN(S):1 Sanyo Denki San Ace 109R1212H101 (Shrouded 25mm, Push Pull added later on for 4.4 and 4.6GHz stops)
Current......................................... 0.52A
Watts............................................ 6.24
RPM ............................................. 2600
Flow CFM (Max)............................ 102.4 CFM / 2.9 cu metre[m3]
Static Pressure (in. H20)................ 64.7Pa (6.6mm)
Noise (dBA).................................... 39.0
CONTROLLER: Sunbeam Rheobus 6Ch.
TIM: AC MX2

WATER:

imgp0638sb.jpg
imgp0764so.jpg


BLOCK: Swiftech GTZ
RAD: XSPC RX360
PUMP: Laing DDC3.2
RES: XSPC Pumptop Res
TUBING: Primochill LRT Black 1/2" I.D. 3/4" O.D.
FAN(S): 3x Sanyo Denki San Ace 109R1212H101 (Shrouded 25mm).
CONTROLLER: Sunbeam Rheobus 6Ch.
TIM: AC MX2



Test Methodology:

  • Each cooling setup will be tested from 6v to 12v fan settings in 2V increments. That is, from minimum sustainable start up speed all the way up to 12V.
    Actual voltages and increments are as close as can be approximated with an analog dial.
  • Each cooling setup will be tested for 10 minutes at each fan speed with LinX (Problem Size 14135). After which, Core Temperatures will be recorded with Everest 5.02.
    In my experience with this loop, things equalize after about 10 minutes of full load. Assuming no change in ambient, there is no noticeable change in load temps from 10 minutes to 20 minutes. And if there's no change after 20 minutes, chances are good that things are equalized.
    Individual core temperatures can be seen in each screencap. For graphical purposes, temperatures have been averaged. I'll warm things up prior to the 6V test to create the same conditions the rest of the fan setting tests will see.
  • Each cooling setup will be tested at the same ambient air temperature.
    This will be somewhere between 20 and 22degC depending on the day, but both water and air will be tested at the same ambient air temperature at each fan speed.
  • Each cooling setup will be tested at 4 different Frequency/Vcore combinations in order from least to greatest. 4GHz, 4.2GHz, 4.4GHz and 4.6GHz.
    This will take a few days, so I'll start with 4GHz and the thread will be updated with the rest as RL allows.

Table of Contents:
Main body of testing: Post 1-5
Air Water testing: Post 49


Enjoy!
 
Last edited:
4.0GHz

AIR

Noise Results

6V - Just Bearing noise. Fans barely audible. Definitely bearable for 24/7 use in living space.
8V - Slight Bearing noise and slight fan noise. Bearable for part time use in living space.
10V - Bearing noise and loud air. Not bearable for use in a living space.
12V - Bearing whine and loud air. Not bearable for use in a living space.

Temperature Results

Thermometer Shot


--------------6v---------------------------8v----------------------------10v-------------------------12v-----------


WATER

Noise Results

6V - Slight Bearing noise and slight fan noise. Just bearable for 24/7 use in living space.
8V - Slight Bearing noise and fan noise. Bearable for part time use in living space.
10V - Bearing noise and loud air. Not bearable for use in a living space.
12V - Bearing whine and loud air. Not bearable for use in a living space.

Temperature Results

Thermometer Shot

--------------6v---------------------------8v----------------------------10v-------------------------12v-----------

OVERALL

watervstruebar.gif


Observations:

1. Alright, as expected, from a Noise/Thermal Efficiency standpoint, the Water wins out across the board. At the 6V level, the Water absolutely whips the TRUE. The Denki's were just not able to effectively penetrate the tightly packed fins of the TRUE at such low speeds. However, as fan speed and static pressure increase, the TRUE comes into its own VERY quickly. It only took a small increase to the 8V level, b4 the TRUE yielded an impressive load temperature of 72.5degC at a noise level that was definitely bearable for use in a living space.

2. With these fans, the 8V level ( or say 1200-1500RPM, 50-70cfm,20-30dba and highest Static Pressure rating you can find) is definitely where you want to be with the TRUE from a noise/performance standpoint. 6V with the water. Assuming your ambient room temperatures don't fluctuate a lot, and your machine specs stay the same over time, increasing the fan speed on either cooling solution nets very little in the way of core temperature decrease. We're talking less than 1.5 degC from 8v-12v on both air and water. I might as well mention, the warmer exhaust air from the rad did not increase the ambient air temperature around the machine in the slightest. That is, the rad fans did not appear to be intaking increasingly warm air. A diverter was used to sweep air out from under the machine and into an open air section of the room.

3. Whether 8V on Air or 6V on water, the noise level is almost identical. In effect that means that with a AIR cooler that cost a total of $85 CDN from the local store, I am seeing temperatures that a mere ~6deg warmer than with a WATER cooler that cost well over $400.00 CDN with shipping and exchange. Hmmmm. Food for thought?

I said it b4 and I'll say it again: If you're looking to get into water cooling as a hobby, power to you. It's friggin awesome, and will likely allow you to OC a little further, and run your system at full load across a broader range of ambient temperatures. Namely warmer summer days.
If you're looking to water cool your rig because you heard it will allow you to dramatically increase your overclocks, please look at the cost/performance comparison above, save your money and spend some more time on your voltage and bios tweaking skills instead. That's good advice.
One should mention that this does not go for all air coolers. There are good and there are bad. The TRUE is the best I've used so far and definitely tames an overclocked i7.
 
Last edited:
4.2GHz

AIR

Noise Results

Same as 4.0GHz stage.

Temperature Results

Thermometer Shot

--------------6v---------------------------8v----------------------------10v-------------------------12v-----------


WATER

Noise Results

Same as 4.0GHz stage.

Temperature Results

Thermometer Shot

--------------6v---------------------------8v----------------------------10v-------------------------12v-----------

OVERALL

truevswaterbar42a.gif
 
Last edited:
4.4GHz

AIR

Noise Results

Addition of PULL FAN changed the noise levels from the 4.0 and 4.2GHz stops somewhat:

6V - No Bearing noise. Slight fan noise. Definitely bearable for 24/7 use in living space.
8V - Slight Bearing noise and moderate fan noise. Bearable for part time use in living space.
10V - Bearing noise and loud air. Not bearable for use in a living space.
12V - Bearing whine and loud air. Not bearable for use in a living space.

Temperature Results

Thermometer Shot

----------10v---------------------12v--------


WATER

Noise Results

Same as 4.0GHz stage.

Temperature Results

Thermometer Shot

----------6v---------------------8v------------------------10v-------------------12v-----------
OVERALL

44htdis.gif


Observations:

  • OK, as expected, 4.4GHz @ 1.4v with HT enabled just doesn't work so well.
    Only the Water Cooled system would keep the cores cool enough to return decent results throughout the fan voltage range.
    Only the 10v and 12v settings on AIR were able to keep the system stable enough to pull off a 10min test, and actually returned decent (albeit somewhat loud) results.
    The 6v and 8v AIR tests returned errors in LinX, which could not be remedied without increasing vcore, which in turn increased temperatures to the point where BSOD's started to occur.
  • Hence HT was disabled and LinX run with 4 threads for the 4.4 and 4.6GHz tests.
  • Another thing I'd like to mention about the 4.4 and 4.6GHz stops, was that I added a PULL FAN to the air setup. Without the pull fan, LinX returned errors without exeption.
    With the pull fan, the 10v and 12v tests of the 4.4GHz stop returned decent results.
  • Although 4.4 and 4.6GHz are doable, 4.2 GHz seems to be the maximum frequency where you can expect any measure of acceptable cooling and 24/7 stability with HT enabled (AIR or H2O). If you want acceptable noise level to go along with that, the 4.0GHz level is where you want to be (AIR or H2O).
 
Last edited:
4.6GHz

AIR

Noise Results

Addition of PULL FAN changed the noise levels from the 4.0 and 4.2GHz stops somewhat:

6V - No Bearing noise. Slight fan noise. Definitely bearable for 24/7 use in living space.
8V - Slight Bearing noise and moderate fan noise. Bearable for part time use in living space.
10V - Bearing noise and loud air. Not bearable for use in a living space.
12V - Bearing whine and loud air. Not bearable for use in a living space.


Temperature Results

Thermometer Shot
----Thermal Shutdown------Thermal Shutdown---------------Unstable-------------------Unstable---------
--------------6v---------------------------8v----------------------------10v-------------------------12v-----------


WATER

Noise Results

Same as 4.0GHz stage.

Temperature Results

Thermometer Shot


--------------6v---------------------------8v----------------------------10v-------------------------12v-----------

OVERALL

46htdis.gif
 
Last edited:
:( RealTemp is THE tool for Intel CPU's (especially for Nehalem/Lynnfield). Or if you don't like it CoreTemp is OK too. Why those two? Because. :)
Otherwise thanks a lot for your tests, very-very informative and interesting. :thup:

Ahhh, good ol' RealTemp. Yes burebista, you and I have done enough posting in the same threads, that you should know I'm a fan of webb's work. :) You can put Everest on the safe list there.
As with 775, RT/CT/Everest read the same within a degree or two at any given time whether at load or idle. Whenever I DL a new version of Everest, RT is the program I use to validate the temperatures being output. Once I am satisfied, RT usually goes back in the drawer.
I simply use Everest because I like to have an eye on more than just my CPU core temperatures and I like the functionality and GUI.
For the sake of the thread, here are a few comparison screen shots. One idle and one load with LinX.

IDLE-Ambient 25degC.
rtctevidle.png


LOAD 1min - Ambient 25degC.
rtctevload.png


LOAD 15mins - Ambient 25degC.
rtctevload15.png


But I'll tell you what, I'll include a mini mode version of RT in the rest of my testing. :)
 
Last edited:
[...] you and I have done enough posting in the same threads, that you should know I'm a fan of webb's work. :)
Yep, I know so that's why I was puzzled when I saw Everest instead RealTemp. :)
Just for my curiosity, in taskbar your temp are white=RealTemp, aqua=CoreTemp and maroon=Everest?
 
Yep. That's right. The core orders appear to be different between the three, even though RT and everest are going 1,2,3,4 and 0,1,2,3.
 
Not only cores order. If you look in your first 2 screenshots it looks like Everest is just duplicating 2 cores temperature.
So from your screenshots I understand that RealTemp always shows correct temps and cores order, CoreTemp always show correct temps and shuffle cores order and Everest sometimes shows correct temps (third screenshot) and sometimes it just duplicate 2 cores plus shuffles cores order.
That's why I said that RealTemp is THE tool when it comes to Intel CPU's. :)

PS. In RealTemp Settings window you can see APIC ID. That is you core order seen by your OS and it can be different at every boot. That's why some monitoring software shows cores shuffled.
 
Really nicely done. Looking forward to more results. An idea for further testing would be another fan brand/model.

This is a straight up comparison between air and water through a Frequency and voltage range. I said right off the bat (top of first post) that I would only test my current setups, in the way that they were currently configured.
I just don't have the time, energy, money or inclination to switch out the fans and go through the whole process again.
It's up to the reader to try and find a fan that would come close to replicating the Denki(or better) at any given voltage value, if so inclined.
Take the results from this thread, and go over to XS or Martins(Skinnee's?) to see if you can find some fans that would give you the best chance at low noise air, as an alternative to water.
http://martinsliquidlab.i4memory.com/FanTesting.html
http://www.xtremesystems.org/Forums/showthread.php?t=171661
http://www.xtremesystems.org/Forums/showthread.php?t=170224

Not only cores order. If you look in your first 2 screenshots it looks like Everest is just duplicating 2 cores temperature.
So from your screenshots I understand that RealTemp always shows correct temps and cores order, CoreTemp always show correct temps and shuffle cores order and Everest sometimes shows correct temps (third screenshot) and sometimes it just duplicate 2 cores plus shuffles cores order.
That's why I said that RealTemp is THE tool when it comes to Intel CPU's. :)

PS. In RealTemp Settings window you can see APIC ID. That is you core order seen by your OS and it can be different at every boot. That's why some monitoring software shows cores shuffled.
I went into the settings tab in APIC ID to give you that information about the Core order (1,2,3,4).
If there were one thing I could change about everest, it would be the apparent core temp replication at times, but for the vast majority of screen captures in this thread, they do not replicate. But, I gotta be honest with ya, it's just not a big enough difference to be even remotely worried about.

Anyhow, thx for the replies, but we're going off track and I really don't want the thread to go down this path. Everest is sufficient to read the temperature of the cores on an i7 and more than sufficient for the purposes of this real world comparison.
If anyone sees a fundamental flaw in my setup or methodology, please fire me a PM and I'll be happy to address it.
 
Last edited:
Nice fans baditude_df , i have one full power in my case cant hear it over the AC :) , these results are just as i expected , thank for doing the work .
 
This is really a very informative and usefull review, always wanted to see some real data to determine the cooling effectiveness water has over air, well done and thanks :)
 
Hmmmm, tells me what I need to know. AC has improved to the point where you don't really lose out that much vs water. Besides, many of us (well, me anyways) are just a little scared of having water in something electronic. Maintenance becomes more important instead of a set and forget type thing.
 
Back