PerformanceAfter the initial installation the waterblock was immediately removed and the contact area was inspected. After loosening the screws that hold down the block, there was a suction between the block and the chipset, but with a very little twist it came off
The fact that the thermal paste was spread out is an indication of good contact.
For the overclocking tests, I used the latest version of SysTool.
The thermal diode integrated in the nForce4 core was used to obtain temperatures, temperatures were read using Systool. Load temperatures were measured after one hour of SysTools torture test, idle temperatures after one hour of Windows desktop.
The DFI motherboard’s chipset varies its fan speed, I set it to 100% for all air tests.
The waterblock was tested with the following water loop:
- Eheim HPPS Plus
- Black Ice Xtreme II with 2 Sunon 7W fans
- Alphacool Nexxxos CPU block
- Thermaltake W3 block
Because I used Arctic Silver Ceramique, I gave the paste three days of gaming in order for it to settle and achieve maximum performance.
I also lowered the LDT and CPU multiplier, put the memory on a divider, and raised the CPU voltage to make sure that I would not be limited by the CPU or memory.
|nForce 4 SLI||Maximum Clock||Temperature Load||Temperature Idle|
|Stock cooler, 1.6V||347 MHz||46°C||38°C|
|Stock cooler, 1.7V||353 MHz||51°C||44°C|
|Stock cooler, 1.8V||358 MHz||55°C||46°C|
|Thermaltake W3, 1.6V||360 MHz||40°C||40°C|
|Thermaltake W3, 1.7V||365 MHz||41°C||40°C|
|Thermaltake W3, 1.8V||366 MHz||43°C||40°C|
I was also interested in the effect that adding a chipset block would have on CPU temperatures. Interestingly, the rise in CPU temperatures was minimal – only about 3°C.
The graphs above show that the W3 helps lower chipset temperatures. It is interesting to see that the stock air cooler provides better cooling at the standard voltage. This might be because it has better contact, or because the chipset sensor is reporting inaccurate temperatures.
I was also interested to see the restriction of this waterblock. In order to obtain flow readings, I used a Siemens Megatron 2 flowmeter. While its accuracy is not ideal, it does give some sort of idea about flow rate.
|Flow without block||0.182 m³/h|
|Flow with block||0.087 m³/h|
From the results we can see that the block decreases water flow by about 50%. I was surprised to see this result, I thought that due to the size and tubing used, the block would be much more restrictive.