Next: GUI platform comparison
Up: GUI energy characterization
Previous: Color patterns
From the application perspective, a user interacts with GUIs
through different input methods, and uses a certain input method
to interact with a certain type of window.
Table XI provides the energy consumption for
different input methods on the targeted handheld computers. It
also shows the number of stylus taps an input method is equal to
(denoted by #) with regard to energy consumption. The stylus tap
is the most commonly used input method. Hardware buttons are used
to trigger the most-used applications. Virtual keyboard is
necessary for text input (Zaurus also comes with a mini hardware
keyboard we do not characterize in this work). Stylus move is
typically used to move a window like a scroll bar.
Table XI:
Additional energy for different input methods
Input method |
Qt |
Windows |
X/GTK |
|
EU |
# |
EU |
# |
EU |
# |
Stylus tap |
1,100 |
1 |
300 |
1 |
900 |
1 |
Hardware button |
1,400 |
1.3 |
560 |
1.9 |
1,300 |
1.4 |
Virtual keyboard |
4,000 |
3.6 |
4,700 |
15.7 |
1,200 |
1.3 |
Stylus move |
13,500 |
12.3 |
5,700 |
19.0 |
3,000 |
3.3 |
A stylus move is very expensive in all three systems. Moreover, it
is usually associated with continuous screen changes such as
window moving and resizing, for which the energy cost is
significant. Moreover, a virtual keyboard is very expensive on
iPAQ1 and Zaurus because they have the auto-completion feature in
order to accelerate user input. Since reducing usage time saves a
significant amount of energy, auto-completion is generally more
energy-efficient. However, inputting text is much slower using a
virtual keyboard than a real keyboard. Thus, it increases usage
time significantly on handheld computers, and leads to more energy
consumption from the application perspective. For energy
efficiency, stylus move and text input should be minimized.
Next: GUI platform comparison
Up: GUI energy characterization
Previous: Color patterns
Lin Zhong
2003-10-13