X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/book_gpu.git/blobdiff_plain/ed20d271e0c4a028423b481170c4d31311494721..fa1939d2294b408e8a62f2d91149f369f8710113:/BookGPU/Chapters/chapter17/ch17.tex?ds=sidebyside diff --git a/BookGPU/Chapters/chapter17/ch17.tex b/BookGPU/Chapters/chapter17/ch17.tex index 9b1c611..d6e3a31 100755 --- a/BookGPU/Chapters/chapter17/ch17.tex +++ b/BookGPU/Chapters/chapter17/ch17.tex @@ -1,8 +1,8 @@ -\chapterauthor{Guillaume Laville}{Femto-ST Institute, University of Franche-Comte, France} -\chapterauthor{Christophe Lang}{Femto-ST Institute, University of Franche-Comte, France} -\chapterauthor{Kamel Mazouzi}{Femto-ST Institute, University of Franche-Comte, France} -\chapterauthor{Nicolas Marilleau}{Femto-ST Institute, University of Franche-Comte, France} -\chapterauthor{Bénédicte Herrmann}{Femto-ST Institute, University of Franche-Comte, France} +\chapterauthor{Guillaume Laville, Christophe Lang, Bénédicte Herrmann and Laurent Philippe}{Femto-ST Institute, University of Franche-Comte, France} +%\chapterauthor{Christophe Lang}{Femto-ST Institute, University of Franche-Comte, France} +\chapterauthor{Kamel Mazouzi}{Franche-Comte Computing Center, University of Franche-Comte, France} +\chapterauthor{Nicolas Marilleau}{UMMISCO, Institut de Recherche pour le Developpement (IRD), France} +%\chapterauthor{Bénédicte Herrmann}{Femto-ST Institute, University of Franche-Comte, France} %\chapterauthor{Laurent Philippe}{Femto-ST Institute, University of Franche-Comte, France} \newlength\mylen @@ -11,7 +11,7 @@ \setlength\hangindent{\mylen}% \hspace*{\mylen}#1\\} -\chapter{Implementing MAS on GPU} +\chapter{Implementing Multi-Agent Systems on GPU} \label{chapter17} @@ -47,7 +47,7 @@ solution to increase simulation performance but Graphical Processing Units (GPU) are also a promising technology with an attractive performance/cost ratio. -Conceptually a MAS is a distributed system as it favors the definition +Conceptually a MAS\index{Multi-Agent System} is a distributed system as it favors the definition and description of large sets of individuals, the agents, that can be run in parallel. As a large set of agents could have the same behavior a SIMD model should fit the simulation execution. Most of the @@ -299,7 +299,7 @@ collembolas in fields and forests. It is based on a diffusion algorithm which illustrates the case of agents with a simple behavior and few synchronization problems. -\subsection{The Collembola model} +\subsection{The Collembola model\index{Collembola model}} \label{ch17:subsec:collembolamodel} The Collembola model is an example of multi-agent system using GIS (Geographical Information System) @@ -374,6 +374,7 @@ the total numbers of access needed to updated those populations. %\lstinputlisting[language=C,caption=Collembola OpenCL %kernels,label=fig:collem_kernels]{Chapters/chapter17/code/collem_kernels.cl} +\pagebreak \lstinputlisting[caption=Collembola OpenCL Diffusion kernel,label=ch17:listing:collembola-diffuse]{Chapters/chapter17/code/collem_kernel_diffuse.cl} The reproduction, diffusion and culling steps are implemented on GPU @@ -646,6 +647,7 @@ implementation by enforcing a quasi-sequential execution. It is necessary, in the case of MIOR as well as for other ABM, to ensure that each work-item is not too constrained in its execution. +\pagebreak \lstinputlisting[caption=Main MIOR kernel,label=ch17:listing:mior_kernels]{Chapters/chapter17/code/mior_kernels.cl} From the sequential algorithm 1 where all the agents share the same