5333058741
Introduces a intermediate derivation that can be imported into the report derivation, which contains a string cmd that expands the fig variable as needed.
32 lines
993 B
TeX
32 lines
993 B
TeX
\documentclass{article}
|
|
\usepackage{graphicx}
|
|
|
|
\begin{document}
|
|
|
|
\title{Example of Nix + \LaTeX{}}
|
|
\author{Rodrigo Arias Mallo}
|
|
|
|
\maketitle
|
|
|
|
\section{Nbody}
|
|
The nbody program has been executed with varying block sizes while the execution
|
|
time $t$ is measured, as shown in the figure \ref{fig:nbody.test}.
|
|
%
|
|
\begin{figure}[h]
|
|
\centering
|
|
\includegraphics[width=0.45\textwidth]{@fig.nbody.baseline@/scatter.png}
|
|
\includegraphics[width=0.45\textwidth]{@fig.nbody.baseline@/box.png}
|
|
\caption{Nbody times with varying block size}
|
|
\label{fig:nbody.test}
|
|
\end{figure}
|
|
%
|
|
The normalized time $\hat t$ is computed with the median time $t_m$ using $ \hat
|
|
t = t / t_{m} - 1 $. It can be observed that the normalized times exceed the
|
|
maximum allowed interval in most cases, except with the largest block sizes.
|
|
|
|
Once the experiment \texttt{exp.nbody.test} changes, the hash of the experiment
|
|
program will change, therefore the plot will be updated and, lastly, this
|
|
report.
|
|
|
|
\end{document}
|