Next: Window properties
Up: GUI energy characterization
Previous: Event handling and basic
Table VI also shows the additional energy required
for using different types of windows and showing an eighty-letter
text. It is obvious that using different window types consumes
quite different amounts of energy even when the window sizes and
colors are similar, as in a message box and a four-item menu
window. Different window types necessitate different user
interactions too, which further differentiate their energy
consumption. There are several observations worth noting, as
discussed next.
First, the same interactive function may be implemented using
different window types with different energy efficiencies. For
instance, both tabbed panels and a scroll bar can be used to
browse a long list. Their energy consumption differs drastically
since a scroll bar requires many more screen updates than tabbed
panels.
Second, ``Message box" of Windows outperforms others simply
because it allows a ``Message box" to be much simpler than Qt or
X/GTK does. For example, ``Message box" of Qt has 3D effects and
must contain at least one button. This indicates that an
inflexibility in the GUI platform can cause extra energy
consumption.
Finally, Windows performs much worse for ``Menu," although it does
well in other window types, because a menu window is animated in
Windows. When a menu is tapped, the menu window gradually, but
quickly, drops out. Such an animation or continuous screen change
requires many more processor cycles, framebuffer updates and
screen refreshes, thus leading to more additional energy. Using a
scroll bar to browse a window also requires continuous screen
changes. For energy efficiency, such continuous screen changes
should be avoided, at the expense of a slight sacrifice in GUI
aesthetics.
Next: Window properties
Up: GUI energy characterization
Previous: Event handling and basic
Lin Zhong
2003-10-13