Part 3. Test Results (temperature, noise level, thermal resistance)
Our updated testing technique for coolers for Socket A processors was briefly
described in the GlacialTech
Igloo 2310 and Igloo 2400 Review (general issues and the original version
of the technique are given in the Cooler
comparison testing technique). Almost the same approach is used for the coolers
for Socket 478.
We chose the Fujitsu Siemens Computers D1337 mainboard as a test platform.
This board, like the D1289, is equipped with a high-quality temperature
monitoring system based on the NE1617A chip from Philips Semiconductors.
This chip is located very close to the socket which allows eliminating
various errors and distortions and improving quality of measurements which
is required for objective comparison of thermal efficiency of coolers.
To reduce test time we removed the Typical User Medium mode; the CPUBurn
mode includes 4 basic test parts (each lasting 2-3 hours); the ambient
temperature is maintained at 34-35°C (this is a typical temperature for
cases with an optimized vent system).
Testbed:
-
Fujitsu Siemens Computers D1337 mainboard
-
Intel Pentium 4 Willamette 1.9 GHz processor
-
Microsoft Windows XP
The burnp6 utility from the CPUBurn packet is used to get a thermal load
close to the maximum one, and the System Guard utility from Fijitsu Siemens
Computers is used for temperature control.
Diagram 1. Temperature results
Notes
Each cooler was tested with its own rated thermal interface
The diagram shows the combined results
Today we supplement the review with one more characteristic - θja
(thermal resistance junction-to-ambient). The necessity appeared long ago,
but we have been were looking for a way of measuring thermal power generated
by a certain processor sample in real test conditions. Finally, we have
solved the problem, and now we can get true temperature results of the
tested systems and objective data on thermal resistance of these systems.
Diagram 2. Thermal performance
Notes
Thermal resistance θja can be found from the following
equation: θja = (Tj - Ta)/Ph,
where Tj is a core temperature, Ta is an ambient
temperature (here it is 35°C), Ph is processor thermal power
(here it is 67 W).
And now the noise measurements (the testing technique is given in the Cooler
Noise Characteristics and Testing Technique).
Diagram 3. Noise characteristic
Note: The background noise level is 25 dBA
Now let's turn to our Table of Ranks - an integrated assessment of consumer qualities
of the coolers in question.
|
|
 |
|
|
|
