Test Methodology and Results:In my experience the component having the greatest impact on temperature in almost every H20 system is the radiator. A more semantically astute description is the
heat-exchanger. This is where the
rubber meets the road.... How and where your radiator/fan compliment is mounted on the H20 loop will have the largest effect on the item to be cooled. To keep temperatures as close to ambient as possible I have done two things. I keep the system "open" to eliminate heat build-up which occurs in sealed cases and I've chosen a triple radiator and high CFM fan combination. Removed the side panels from the already "breathable" Mozart Tx to eliminate temp variables. Ultimately in a typical H20 system temps cannot be lower then the air-temp at the radiator inlet.
Intel Test System: |
Processor | Intel Q6600 SLACR Socket-775 |
Mainboard | Gigabyte GA-P35C-DS3R |
Memory | Super Talent DDR3-1600 (2x1GB) GEIL DDR2-800 (2x1GB) |
Graphics | BFG 8800Ultra |
Storage | Seagate Barracuda 80GB SATA Perpendicular |
Optical | Plextor PX755-SA DVD/RW |
Power Supply | Tuniq Ensemble 1200W |
Cooling System | Laing Vario D38, Sunon 3x120mm @ 98CFM, Black Ice Xtreme III (triple radiator), Fluid (Distilled H20, Water Wetter, Anti Algae) 1/2" Tygon tubing 1.2m L
|
Original Waterblocks Tested | D-Tek Fuzion Swiftech Apogee GTX Swiftech Apogee GT EnzoTech Sapphire SCX-1 Koolance CPU-330 Danger Den TDX (original) |
Waterblocks Added | Swiftech Storm EK Supreme Koolance CPU-340 |
Operating System | Windows XP |
Chassis | Thermaltake Mozart TX (open chassis) |
Testing MethodologyIn the original Round-Up I found some blocks included a "back-plate" with their hardware while others did not. And, as I mentioned on the Koolance page, their mounting hardware seemed quite strange to me. Once that proverbial can of worms was open I chose to run an additional test in which I used Swiftech's back-plate design and applied it to all blocks. The results showed an improvement in performance across the board and also showed CPU-330 to perform better with a back-plate which places pressure on the area directly behind the socket, compared to their own back-plate which encircles it. In the first two charts each water-block was tested with the hardware supplied, in all fairness most people would not be modifying nor installing a modified "back-plate." In the third chart I employed a modified Swiftech style back-plate which exerts pressure behind the Socket. At the time of writing about all the blocks tested now offer back-plate kits. Therefore the first two charts below represent CPU temps with water-blocks running on number #3 and #5 settings on the D5-38 Vario pump, or the difference between pressure/flow rate.
Each water-block was mounted tested and re-mounted at least three times to ensure thermal paste was distributed evenly and the contact between surfaces was consistent. Out of three (or more) installations the lowest consistent temp was chosen for the final result (chart). Thermal paste remains unchanged from the last review;
Tuniq TX2 which sets-up quite rapidly. I used Arctic Silver's
AS5 Intel Quad Core application method which basically involves laying down two lines perpendicularly across the cores forming an "X" and distributing the paste to each core.
Ambient temp for all tests remained between
18°C ~ 19°C measured at the radiator inlet.
As mentioned above the chart below tests the blocks running on the # 5 setting and including a back-plate exerting pressure behind the Socket. Not only did this improve performance across the board it improved performance on the Koolance water blocks even compared to their own back-plate design.
ConclusionFirst let's address the Koolance CPU-340. How a Nickel plated copper base out-performs a Gold plated copper base which are "almost" identical perplexed me, until I realized the CPU-340 simply has more mass. While the CPU-330 is "larger" it weighs less. Without an accurate scale it's difficult to tell, Koolance states the CPU-330 weighs in at 198g, in my estimation the CPU-340 weighs in at 250g.
Second, the EK Supreme, this block bested all the others including the D-Tek by a few degrees, at the highest pump setting. Although on the Laing D5-38 Vario's #3 Settings the Fuzion reigned (over the) Supreme. I attribute this to the EK Supreme's reliance on a high pressure pump for ultimate performance.
Koolance CPU-340: at $54.95 MSRP is just $10 more then it's sibling the CPU-330 and worth the price increase there. The back-plate hardware should be re-designed.
PRO
decent performance
all metal design
nickel plated
CON
mounting hardware may be defeatist
priceEK Supreme Plexi: retails for $69.95 USA (45.95 EUR). The sample I received didn't include a back-plate however; these are now available.
PRO
best performer!
CON
should include back-plate for the price
price
We thank Koolance and EK for allowing us to test their product, and also Swiftech for sending us the Storm water block!