Test Setup and Methodology
Geoffreys' Intel Test Setup
|CPU||Intel Q6600 G0 L724 @ 2,4GHz (1,22V)Intel Q6600 G0 L724 @ 3GHz (1,52V)|
|Cooling||Intel stock heatsinkZalman 9500 LED heatsinkTTIC Micro Flow waterblock|
|Mainboard||ASUS P5K (vanilla)|
|Memory||2x1Gb OCZ Reaper PC2 8500 5-5-5-15|
|Other||Silverstone DA750 / EPS Epsilon 900 PSUMaxtor 80Gb PATA HDD20" Dell UltraSharp 2007FP TFT monitorAntec 900 housing|
The goal of our review is to inform our readers with what they could expect when purchasing a TTIC Micro Flow water block. The best way to do this was to compare this block with Intel's boxed heatsink, this way we would have a clear view on how the Micro Flow could be a possible upgrade path for those who want to get into H20 cooling. On the other hand, high-end heatsinks have upped the ante recently, they are so powerful that the H20 cooling market certainly may have noticed lower sales rates and therefore we added our Zalman 9500 LED heatsink in this article.
Recent technology has bought us CPU's with 4 physical cores in one tiny package, combined with the CPU price war we found that the Intel Q6600 is quite an excellent bargain, even though most software doesn't really use all 4 cores. At stock settings this CPU doesn't dissipate as much heat as you might suspect, but once you start cranking up the voltage and MHz's you'll notice how this processor is in desperate need of advanced cooling equipment. I won't go into details yet, for now I'll just inform you that we tried two different clock settings and that the difference between them is... noticeable.
The CPU was tested at stock settings, being 2,4GHz (266*9), real Vcore as measured on the board and was around 1,22V when the CPU was heavy stressed. With the overclocked profile processor running at 3GHz by setting the front side bus to 378MHz and keeping the multiplier at 8, the real CPU Vcore measured was 1,52V. The memory was running @ 378 MHz (756MHz DDR) with 3-3-3-8 1T timings, 1/1 with the FSB.
Last but not least, water-cooling performance is very dependant on which components are used in the water loop. Pump, tubing size, radiator design and cooling surface, reservoir size, fan airflow, liquid used, ... results obtained in this review can vary much from those you're getting back at your place, here is a picture of our water loop:
Geoffreys' Watercooling Test Setup
|Type||External, no housing|
|Radiator||Black Ice Extreme II|
|Pump||Laing D4 12V|
|Reservoir||Selfmade, 1/2" fittings|
|Tubing||Clear PVC 1/2" ID|
|Fans||Enermax B12025112M-3M 12V 120mm fan 12VPapst 12V 120mm fan, unknown type|
For this article we tried two different voltage levels for both fan and pump, this way we'll have a good view on how the block will perform in different setups. We tested them at 12V and at reduced duty cycle being 7V.
4 physical cores means that we have to use multithreaded software in order to properly stress a Quad Core Q6600 CPU. wPrime
is an excellent tool to push the CPU to 100% load, we used its 1024M test in loop for 1/2h and measured the CPU core temperature with Core Temp version 0.95.4, a tool which reads the temperature for each different core. In real life the temperature between the CPU cores, IDLE or LOAD, always differs a bit, therefore I averaged the temperature so that our temperature charts become easier to read.Installation
Our sample didn't have a manual; nevertheless, the install procedure is straight forward and went without any troubles. The block is tightened by using 4 bolts with pre-applied springs; this means that you do have to remove your mainboard from the case before you can get the block mounted on your CPU. The bolts are screwed in a back plate which has a central square pad to make sure that the block is pressured onto the CPU without bending the mainboard. Getting the right mounting pressure isn’t a problem either, you screw the bolts only to a certain point, that way you won't over tighten the block causing anything to happen to your mainboard or processor.
Removing the water block went less flawless. The copper block uninstalled easily, but the sticky square pad at the back of the mainboard was way to strong and we had to pull very hard before it wanted to loosen. It was even so that we pulled of the back plate itself first leaving the square pad still hanging on the motherboard, afterwards the sticky pad came of rather easily though over time it will certainly loose its bounding capability and might even get damaged after multiple installs. We hope TTIC well add more pads or find a different material to stick on their back plates, that would make the installation near perfect.
Onto the performance chart, read on ->