CPU: | AMD Athlon64 3000+ (S939; 512 KB; Venice) |
---|
Clock speed: | 9 x 200 MHz = 1800 MHz, Memory at DDR-400 |
---|
Motherboard: | ATI RX480 Reference Board ATI RX480 |
---|
Memory: | 2x 256MB Generic PC-3200 |
---|
Video Card: | ATI Radeon 9000 PCI |
---|
Harddisk: | WD Raptor 360GD 36GB |
---|
Power Supply: | HEC Power475 |
---|
Software: | Windows XP SP2, Catalyst 6.1 |
---|
Temperature data was obtained by directly reading the Athlon64 internal thermal diode. It offers an accuracy of about 1-2°C.
Idle means Windows sitting at the desktop for 30 minutes. Load is after 30 minutes of Prime95. The different clock/voltage settings are representative of an un-overclocked, medium overclocked and heavily overclocked system. The high voltage settings were used to get a sufficient heat output on the CPU, not to reach the overclock.
Very impressive, the Noctua NH-U12F can take the lead of all the heatsinks tested here. Especially when heat load is not extremely high it gains quite a nice margin compared to the other heatsinks.
What I did notice during testing is that it helps temperatures a lot if you release the screws about half a turn after locking them down all the way.
Fan Noise
To measure fan noise we used an IEC Type 2 sound level meter on the dbA setting. Measuring distance was 10 cm from the heatsink fan hub. The short distance of 10 cm is necessary to get proper readings with very silent fans. All fans were tested outside of the case at 12V supplied by a lab PSU. On fans that come with a fan controller or allow control of fan speed in any other way, "low" and "high" indicate the settings on the fan controller.
While the NH-U12F is not the quietest heatsink, it is quiet enough to be used in any work / living room PC where you can live with a bit of fan noise. If you need an even quieter system you can use the included Ultra Low Noise adapter which runs the fan at a slower speed to make it almost inaudible, but cooling performance is reduced as well.