Testbed and Methods
The tests are performed with a closed and assembled system case at a constant ambient temperature of 23° that is maintained by an air conditioner. Most users are likely to prefer quieter PCs, so we set the CPU cooler and the system fans (connected to the mainboard) into the Silence mode. If the system case has its own controller, the fans connected to it are set at minimum speed, too. We did not modify the configuration of airflows.
The following configuration was assembled in the tested system case:
- ASUS P5G-MX mainboard
- Intel Core 2 Duo E6850 CPU
- Western Digital Raptor WD740GD HDD
- 2GB DDR2-800 SDRAM (Patriot PDC24G6400LLK)
- Microsoft Windows XP Professional SP2
The microATX mainboard I used does not support a system bus frequency of 1333MHz, so the E6850 processor was clocked at 266x9=2400MHz rather than at its standard 3GHz. Anyway, the heat dissipation of this CPU, even at a reduced frequency, is comparable to today’s 45nm dual-core CPUs and higher than that of Atom CPUs. I don’t think that users are going to install a more advanced processor into an HTPC anyway. Today’s entry-level CPUs are more than enough for decoding HD video with support from the integrated graphics core.
The CPU temperature was read with the ASUS PC Probe program supplied with the mainboard. The HDD temperature was reported by HDD Thermometer. The fan speeds were measured with an optical tachometer Velleman DTO2234.
There were the following test modes:
- IOMeter (IOMeter’s Access Time test runs on the HDD, loading it to the full)
- Prime95 (Prime95 is running in the in-place large FFTs mode that loads both CPU cores fully)
- 3DMark06 (3DMark06 is running at 1280x1024 with the highest graphics quality settings)
Each temperature was read after half an hour of operation in each test mode, i.e. when the temperatures of the components had stabilized.
As a reference point, here are the temperatures of a similar configuration working without a system case (“open testbed”).
The system case configuration differed from the open testbed in having no graphics card and being equipped with different coolers (the open testbed was tested with a Zalman CNPS9500 AT).
The level of noise was evaluated subjectively.