Viewing page 4 of 5 pages.
Previous 1 2 3 4 5 Next
TESTING
The EPIA EN12000E was tested for thermal performance and power consumption.
The tests were done with the system in three states: Idle (in Windows), CPU Load (using CPUBurn),
and Full load (using CPUBurn
and ATI Tool
simultaneously to stress both the CPU and graphics). CPU temperature was monitored
using VIA's Flight Deck application, although SpeedFan was also capable of reading
the thermal diode. System power was measured with a Seasonic
Power Angel at the AC plug and does not include the power required by
the monitor.
The rest of the components in the system:
- 512 MB generic DDR2 533 MHz RAM
- Seasonic Super Silencer 300 SS-300FS power supply
- Fujitsu MHT2080BH: 80GB, 5,400 RPM SATA notebook drive
Throughout the testing, the only sources of noise were the hard drive and the
power supply. For this reason, noise was not measured, as the board itself had
no effect on the noise level of the system.
Ambient conditions during testing were 21°C and 121V @ 60 Hz.
|
VIA EPIA EN12000E: Thermal and Power Measurements
|
|
Activity State
|
CPU Temperature
|
AC Power Draw
|
|
Idle
|
26°C
|
28W
|
|
CPUBurn
|
35°C
|
33W
|
|
CPUBurn + ATI Tool
|
35°C
|
35W
|
The results of the test made the EN12000E look very, very good. The idle temperature
was just 5°C above room ambient, and no wonder — the total system power
was just 28W! We have never before managed to get any system to idle
at less than 37W, even with a Pentium M with SpeedStep enabled.
Things did not change much under load. With the CPU under heavy stress, the
system power increased by only 5 watts and the temperature by 9°C to top
out at 35°C. Unfortunately, the thermal data is a little bit difficult to
interpret, since the accuracy of the sensor and the thermal tolerance of the
CPU is unknown. However, assuming reasonable accuracy in the Eden processor
thermal sensor,
a 35°C CPU temperature should have plenty of headroom as ~60°C is usually
considered a reasonable target for mainstream processors. VIA C3 processors have been known to keep running indefinitely even without a heatsink.
IN-SYSTEM TESTING
The only usable intake vent is quite small...
Our open air test bench clearly did not present much of a challenge to the
EPIA, so we decided to see how it did in more strenuous circumstances. An old Sereniti
2000 enclosure, one of the first designed specifically for Mini-ITX systems, was called into
service. The Sereniti has terrible cooling airflow. Two tiny intake vents on either side of the case
are the only sources of air, and airflow is expected to be generated by the
power supply.
We weren't interested in using the stock power supply, which is loud and inefficient,
so we were faced with a problem: How to fit a power supply into the tiny confines
of the our chosen enclosure? We did not have any other power supplies of the
same form factor as the original, but we did have a
picoPSU, which was designed originally for mini-ITX.
The picoPSU is a tiny DC-DC power supply that allows the system to be powered
by an external power brick. As a bonus, it is both fanless and efficient.
An emergency cooling fan was installed where the original power supply was.
In addition to swapping the power supply for a picoPSU, we also took the opportunity
to round out the system by adding a DVD drive — an essential addition for
a home theater. When not in use, the power it consumes is negligible, but it
could become a significant factor when spinning quickly.
The DVD drive was added for the system comparison.
|
VIA EPIA EN12000E: In-System Test
|
|
Activity State
|
CPU
|
System
|
AC Power
|
|
Idle
|
30°C
|
52°C
|
17W
|
|
CPUBurn
|
41°C
|
56°C
|
21W
|
|
CPUBurn + ATI Tool
|
44°C
|
59°C
|
24W
|
This time, SpeedFan
4.28 was used to monitor the temperatures, and — surprise! —
an unknown thermal sensor suddenly showed up. Ordinarily, we do not monitor
sensors when we do not know what they mean, but this sensor was special: It
read much higher than the "CPU" temperature. We decided that it was
worth paying attention to. Further investigation in the BIOS revealed a sensor
labeled "System"; it seems reasonable that SpeedFan was reporting
this sensor.
Even inside the poorly-ventilated Sereniti 2000 with no fans running, the EPIA
was still quite cool, and perfectly stable to boot. Temperatures rose by about
5°C across the board: A tiny amount that is almost insignificant.
Unlike the open-air bench, there was a slight but significant difference between
the CPU-only load and the full CPU/Graphics load. With the graphics load in
play the CPU temperature went up 3°C — most likely because of an increase
in the ambient air around it.
On the other hand, the AC Power dropped dramatically thanks entirely to the
highly efficient PicoPSU. While the 28W idle from before was impressive, the
17W idle with the picoPSU more than doubled the power advantage over the previous
best. At idle, this EPIA system required less than half the power of any other system
we've measured.
The power consumption under load was even
more impressive, topping out at just 24W — still much less than any other
system at idle.
| Help support this site, buy from one of our affiliate retailers! |
|
