The following is reposted from my original article dated 4/2/09
I had previously asked Stephen to conduct an informal set of tests on our scientific bench in order to compare the Apogee GTZ to the Koolance 350AC. Since there appeared to be some disconnect between our tests and user reports on the Xtremesystems forum, I decided to redo the tests myself. Our scientific bench was also commented by some of their members as being too distant from an Xtreme user setup; so I set out to use the “Xtreme User bench” that we have in the lab. I call it Xtreme user, because it is in fact quite close to what many of the top dogs use around here: (2) triple MCR320 rads, and (2) MCP655 pumps (vario at 5). The fans are our standard RDM1225S running at 12v, and specified for 81.3 CFM’s.
Finally, since the HK3.0 recently became available in the US, I decided to also include it in the round-up.
Here is a picture of the “Xtreme” setup:
Note that we use a Meanwell S32012 to power up pumps and fans; that way all we need to do is flip a wall switch to start cooling - The Meanwell is calibrated at 12v. We also use non-spill quick-disconnect fittings. Flow restriction is notable, but convenience is priceless in a lab environment.
To capture the temps we use a Fluke 2180A RTD digital thermometer for the coolant accurate at +/- 0.01 ºC, and a 2190A for air, accurate at +/- 0.1 ºC.
Now to the testing.
System hardware:
Motherboard: Asus Rampage 2 Extreme
CPU: Ci7 920 overclocked to 4095 MHz at 1.392v, Intel Turbo mode enabled, HT enabled
Full-cover water-cooled GTX285 in series with the GPU
Apogee GTZ: picked at random from our inventory
Koolance 350AC: purchased retail.
HK 3.0: purchased retail.
Testing protocol:
3 mounts for each block, using Arctic Lumiere for thermal compound, which is a laboratory grade of Arctic Ceramique.
The systems were allowed to ramp up to stable coolant temperature.
CPU, ambient air and coolant temperatures were recorded every minute on the minute for a period of 10 minutes using RealTemp 3.0
CPU load was applied using 8 threads of BurnK6.
GTZ system setup
350AC system setup
HK3.0 system setup
INSTALLATION:
Apogee GTZ:
Nothing to report other than painless and repeatable.
Koolance 350AC:
User is required to estimate how much spring tension to apply. To make sure that I had an absolutely even mount at all 4 screws, I measured the distance between thumbnut and hold-down plate and made sure that it was identical at all four screws within 0.1mm (0.004”). The tension to the springs was the maximum I could apply with my fingers.
HK3.0
No back plate is supplied with the block. Users are required to precisely measure each spring height, using similar technique to the above; a height table in mm is provided. Recommended spring tension of 20Nm was used for the test.
Here is a summary of the test results. The complete spreadsheet is available for download here: http://www.swiftech.com/ASSETS/IMAGE...,0-results.pdf
Note that all individual CPU temps show 1 decimal, as they each represent the average of 9 measurements. Air and coolant temps are also the average of 9 measurements.
Note: where “∆” stands for Delta, and is meant as “difference”
OBSERVATIONS AND CONCLUSIONS
Given the precision of the CPU DTS (+/- 1ºC) these test results are accurate at +/- 1ºC. We can observe that GTZ and HK3.0 are tied, and that 350AC is visibly trailing behind.
Reply With Quote
*
Originally Posted by Vapor
Bookmarks