Commit ffa568d23fa728eac0444c54bfb8a9a5778aaa5b
1 parent
3ad686ef
New.
Showing
2 changed files
with
27 additions
and
2 deletions
figures/tables/tableALL.tex
| @@ -137,9 +137,9 @@ | @@ -137,9 +137,9 @@ | ||
| 137 | \end{tabularx} | 137 | \end{tabularx} |
| 138 | } | 138 | } |
| 139 | 139 | ||
| 140 | -\label{table-content} | ||
| 141 | - | ||
| 142 | \caption{\small \textbf{Apps sorted by content energy efficiency.}} | 140 | \caption{\small \textbf{Apps sorted by content energy efficiency.}} |
| 141 | + | ||
| 142 | +\label{table-content} | ||
| 143 | \end{subtable} | 143 | \end{subtable} |
| 144 | 144 | ||
| 145 | \caption{\small \textbf{Evaluating Components of a Value Measure.} | 145 | \caption{\small \textbf{Evaluating Components of a Value Measure.} |
results.tex
| @@ -82,6 +82,31 @@ looked better when their foreground usage was considered. | @@ -82,6 +82,31 @@ looked better when their foreground usage was considered. | ||
| 82 | 82 | ||
| 83 | \subsection{Content Energy Efficiency} | 83 | \subsection{Content Energy Efficiency} |
| 84 | 84 | ||
| 85 | +Finally, we the data we collected by instrumenting the | ||
| 86 | +\texttt{SurfaceFlinger} and \texttt{AudioFlinger} components to compute a | ||
| 87 | +simple measure of content delivery. We measure the audio and video frame | ||
| 88 | +rates and combine them into a single measure by using bitrates corresponding | ||
| 89 | +to a 30~fps YouTube-encoded video and 128~kbps two-channel audio, with the | ||
| 90 | +weights representing the fact that a single frame of video contains much more | ||
| 91 | +content than a single sample of audio. We use this combined metric as the | ||
| 92 | +value measure and again use it to weight the energy consumption of each app, | ||
| 93 | +with the results shown in Table~\ref{table-content}. | ||
| 94 | + | ||
| 95 | +Comparing with the foreground energy efficiency again shows several | ||
| 96 | +interesting changes. Yahoo Mail, which foreground energy efficiency marked as | ||
| 97 | +inefficiency, looks more efficient when content delivery is considered. While | ||
| 98 | +it is possible that one \PhoneLab{} participant uses it to read email very | ||
| 99 | +quickly, it may be more likely that it uses a ``spinner'' or other fancy UI | ||
| 100 | +elements that generate artificially high frame rates without delivering much | ||
| 101 | +information. The inability to distinguish between meaningless and meaningful | ||
| 102 | +video frame content is a significant weakness of this simple approach. | ||
| 103 | +YouTube and Candy Crush Saga both earn high marks, which is encouraging given | ||
| 104 | +that they are very different apps but also might be a result of overweighting | ||
| 105 | +screen refreshes. The Android Clock is also an unsurprising result, as it | ||
| 106 | +requires almost no energy to generate a relatively-large number of screen | ||
| 107 | +redraws. | ||
| 108 | + | ||
| 109 | +\subsection{Survey Results and Discussion} | ||
| 85 | 110 | ||
| 86 | \begin{figure*}[t] | 111 | \begin{figure*}[t] |
| 87 | \centering | 112 | \centering |