Viewing page 4 of 6 pages.
Previous 1 2 3 4 5 6 Next
TESTING
Test Setup:
- Intel
Core 2 Duo E7200 processor - 2.53Ghz, 3MB L2 cache, 45nm, 65W
- Apack ZeroTherm BTF90
heatsink/fan
- Asus
P5E3 Premium motherboard - 401 BIOS
- Corsair
XMS3 Dominator TWIN3X memory - 2x1GB, DDR3-1800
- Palit Radeon HD3870
graphics card - 512MB, PCI-E 2.0
- Western Digital WD6400AAKS
hard drive - 640GB, 7200RPM, 16MB cache, 2 platters
- Enermax Modu82+
625W power supply - 625W, ATX12V, 80-Plus
- Microsoft
Windows Vista operating system - Home Premium, 32-bit
- ATI
Catalyst 8.4 graphics driver
Measurement and Analysis Tools
|

Test setup.
|
The components chosen for testing where selected to reflect
a typical, high-end gaming set-up. Except for the CPU and heatsink, they were
the highest performing parts we had on hand. The E7200 processor was not our
first choice, but unfortunately it was the first CPU we tested on the board
that actually worked properly. A QX9650 and E6750 were tried as well, but
for some odd reason, the board automatically overclocked both processors to
450Mhz FSB and increased the CPU voltage, resulting in SpeedStep being disabled
and preventing us from testing the EPU feature, which does not function in
an overclocked state. We attempted to get around it by setting the CPU speed
manually and updated the BIOS, but neither workaround resulted in any kind
of resolution.
EPU
|

EasyTune overclocking.
|
The EPU feature requires a driver as well as the installation of Asus'
AI Gear3+ application. There are five different modes to choose from: Max.
Power Saving, Medium Power Saving, High Performance, Turbo, and Auto. Each
setting is represented visually by different modes of transport, with the
man walking being the slowest and the rocket being the fastest. There is
also an Automatic mode (the wand) that scans the system to determine the
most optimal overclocking settings. When we tried it, it concluded that
the same settings as Medium were appropriate. For testing, we tried each
mode and measured the system power consumption at idle, during playback
fo a VC-1 clip, and during load.
|
EPU Setting
|
CPU Manipulation
|
System Power (AC)
|
|
Mult.
|
FSB
|
Clock Speed
|
CPU Voltage
|
Idle
|
VC-1
|
Load
|
|
Max. Power Saving
|
6.0
|
241Mhz
|
1.44Ghz
|
0.976V
|
76.9W
|
88.3W
|
88.2W
|
|
Med. Power Saving
|
6.0/9.5
|
253Mhz
|
1.52Ghz/
2.40Ghz
|
1.032V/
1.080V
|
80.9W
|
87.9W
|
104.1W
|
|
High Performance
|
6.0/9.5
|
267Mhz
|
1.60Ghz/
2.53Ghz
|
1.128V/
1.176V
|
81.8W
|
89.5W
|
110.1W
|
|
Turbo Performance
|
6.0/9.5
|
281Mhz
|
1.68Ghz/
2.66Ghz
|
1.120V/
1.168V
|
82.2W
|
88.9W
|
111.6W
|
|
CPU voltage reported by CPU-Z.
|
How well EPU performs unfortunately is fogged by the fact that the CPU
clock speed and voltage are also altered. Whether the manipulation of the
CPU's power phases or simple underclocking and undervolting are resulting
in energy savings is hard to tell. Both the idle and load CPU voltage are
changed at each level, effectively altering SpeedStep to suit the clock
speed EPU has set. The changes are rather moderate, with a slight underclock
in effect when using the Medium mode and small overclock using Turbo. Max.
Power Saving is the most drastic, severely underclocked the CPU to 1.44Ghz
(more than 40%) regardless of CPU load. It's fairly easy to reap significant
power savings when performance is crippled.
|
State
|
System Power Consumption (AC)
|
|
Idle
|
H.264
|
VC-1
|
Load
|
|
EPU - High Performance
|
81.8W
|
88.4W
|
89.5W
|
110.1W
|
|
EPU Disabled - Stock EIST
|
81.9W
|
88.8W
|
90.0W
|
110.2W
|
|
CPU speed: 1.60Ghz idle, 2.53Ghz load
CPU voltage: 1.128V idle, 1.176V load
|
So is EPU actually doing anything? After we uninstalled the EPU driver
and AI Gear and rebooted the system, a quick check with CPU-Z confirmed
that the stock settings with EIST enabled were identical to the High Performance
EPU profile. A comparison of the power consumption between the two states
showed very little difference. Our power analyzer rounds off to the nearest
tenth of a watt if it didn't we might not have any results at all. The
power savings were so minimal that they're hardly worth mentioning. It's
quite possible that EPU just didn't work properly on our test sample, as
it did give us trouble early on when we were selecting a CPU to use with
it. Another possibility is the E7200 processor we ended up using is not
power hungry enough for EPU to make any real difference.
| Help support this site, buy the Asus P5E3 Premium/WiFi-AP @n Motherbaord from one of our affiliate retailers! |
|
» 64160 reads