Page 12
Testing and benchmarking
Now that the system is hooked up and running, let's run some performance tests.
First
thought after installation, noisy yet impressive.Power consumption: The power consumption of the pump plus radiator fan activated. A steady 29 Watts will be consumed for cooling. Now with the unit installed let's focus back on how the unit manages temperatures.
System:
- Intel 965P based
- Core 2 Duo Conroe E6600
- GeForce 6800 XT
- 2 GB memory
- 1x 300GB HD
- Silverstone Strider 720 Watt PSU
The CPU was running steady at a ~32 C temperature right after windows booted up yet it climbed up even further. As part of the test we maxed the workload of the CPU to 100% for ten minutes, managing some serious encoding work.
|
Celsius |
Farenheit |
|
20.00 |
68.00 |
|
21.00 |
69.80 |
|
22.00 |
71.60 |
|
23.00 |
73.40 |
|
24.00 |
75.20 |
|
25.00 |
77.00 |
|
26.00 |
78.80 |
|
27.00 |
80.60 |
|
28.00 |
82.40 |
|
29.00 |
84.20 |
|
30.00 |
86.00 |
|
31.00 |
87.80 |
|
32.00 |
89.60 |
|
33.00 |
91.40 |
|
34.00 |
93.20 |
|
35.00 |
95.00 |
|
36.00 |
96.80 |
|
37.00 |
98.60 |
|
38.00 |
100.40 |
|
39.00 |
102.20 |
|
40.00 |
104.00 |
What you need to keep in mind with a water-cooled system is that temperature is extremely dependant on the surrounding ambient heat in your house. So if we measure at 15 Degrees C we'd have completely different results compared to a warm day at 30 Degrees C. So that would have an adverse effect on overclocking as well.
We use the cool tool Core Temp v0.9.0.91 which measures Intel's "Core", "Core 2" and all AMD K8 chips' die temperature. The temperature readings are very accurate as the data is collected from a Digital Thermal Sensor (or DTS) which is located in each individual processing core, near the hottest part.
At 21 Degrees C ambient room temperature the Core 2 Duo processor idles at 32 Degrees C on the first Core and with the second at 31 Degrees C which is quite amazing. But obviously what matters is the temperature when the system CPU is working at 100% for a while.
We loop a CPU intensive encoding job and notice 100% CPU utilization on both CPU cores. Now we monitor for a while and observe what the absolute highest temperature value was.
We have two cores and both are measured.
The CPU at 100% utilization
Fan LOW - Core 1 was maxing out at 43C and Core 2 at 41C.
Fan MID - Core 1 was maxing out at 40C and Core 2 at 39C.
Fan HIGH - Core 1 was maxing out at 39C and Core 2 at 38C.
This was an Core 2 Duo E6600 processor
The VGA Card at 100% utilization
In it's default form the videocard we used (Galaxy 6800 XT) was cooled down pretty darn well as it has a Zalman cooler equipped on it which is very silent. At idle the card with the Zalman cooler was at 53 Degrees C in IDLE and 70 degrees C at 100% CPU utilization. Let's see what the water-cooling will do with it:
Fan LOW - GPU Core was maxing out at 44 C
Fan MID - GPU Core was maxing out at 43 C
Fan HIGH - GPU Core was maxing out at 41 C
I don't know what you think but that's really good for a 150 USD kit ! The temps are really okay. It's almost a high-end high performance unit, mainly due to the tremendous radiator which can dispose heat really well. Again, with a higher room-temperature these results would be higher though.
We overclocked the E6600 CPU with the fan set at high and temps went up at a peak of roughly 42 Degrees C measured over time. Our end result was a ~3.4 GHz E6600 Processor.
Bring on the noise !
Unfortunately there's one negative on the kit. It's noisy. The pump resonates way too much and thus constantly can be heard. Gigabyte really needs to look into this. Sure, it's performance is really good but it does not weigh up against the noise levels.
Next to that the radiator fan is rather noisy as well. On it's lower setting it's okay though, yet you can definitely hear it. The compromise Gigabyte made here was that they have a small radiator, thus you need lot's of airflow to cool down the fluid. If they opted for a bigger radiator then the rotational fan speeds and or size of the fan could have taken away this problem.
All in all at it's lowest setting this system was making a noise level of 44 dBA and at it's highest level it went over 52 DBa.
So overall the kit is looking good and performing extremely well but it's noise levels can clearly be heard. Right let's move onwards to the conclusion.
