X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/book_gpu.git/blobdiff_plain/bdec1b5087c2ea922fcf62ad0591b8d784ddf3b7..1b4c95e988aa2d0c5d7f6cf87750ea4214dc8171:/BookGPU/Chapters/chapter17/ch17.aux?ds=sidebyside diff --git a/BookGPU/Chapters/chapter17/ch17.aux b/BookGPU/Chapters/chapter17/ch17.aux index 032ebff..1236656 100644 --- a/BookGPU/Chapters/chapter17/ch17.aux +++ b/BookGPU/Chapters/chapter17/ch17.aux @@ -1,76 +1,81 @@ \relax -\@writefile{toc}{\author{G. Laville, C. Lang, K. Mazouzi, N. Marilleau, B. Herrmann, L. Philippe}{}} +\@writefile{toc}{\author{Guillaume Laville}{}} +\@writefile{toc}{\author{Christophe Lang}{}} +\@writefile{toc}{\author{Kamel Mazouzi}{}} +\@writefile{toc}{\author{Nicolas Marilleau}{}} +\@writefile{toc}{\author{B\IeC {\'e}n\IeC {\'e}dicte Herrmann}{}} +\@writefile{toc}{\author{Laurent Philippe}{}} \@writefile{loa}{\addvspace {10\p@ }} -\@writefile{toc}{\contentsline {chapter}{\numberline {16}Implementing MAS on GPU}{363}} +\@writefile{toc}{\contentsline {chapter}{\numberline {18}Implementing Multi-Agent Systems on GPU}{417}} \@writefile{lof}{\addvspace {10\p@ }} \@writefile{lot}{\addvspace {10\p@ }} -\newlabel{chapter17}{{16}{363}} -\@writefile{toc}{\contentsline {section}{\numberline {16.1}Introduction}{363}} -\newlabel{ch17:intro}{{16.1}{363}} -\@writefile{toc}{\contentsline {section}{\numberline {16.2}Running Agent-Based Simulations}{365}} -\newlabel{ch17:ABM}{{16.2}{365}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.2.1}Multi-agent systems and parallelism}{365}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.2.2}MAS Implementation on GPU}{366}} -\newlabel{ch17:subsec:gpu}{{16.2.2}{366}} -\@writefile{toc}{\contentsline {section}{\numberline {16.3}A first practical example}{368}} -\newlabel{ch17:sec:1stmodel}{{16.3}{368}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.3.1}The Collembola model}{368}} -\newlabel{ch17:subsec:collembolamodel}{{16.3.1}{368}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.1}{\ignorespaces Evolution algorithm of Collembola model\relax }}{369}} -\newlabel{ch17:fig:collem_algorithm}{{16.1}{369}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.3.2}Collembola Implementation}{369}} -\newlabel{ch17:listing:collembola-diffuse}{{16.1}{369}} -\@writefile{lol}{\contentsline {lstlisting}{\numberline {16.1}Collembola OpenCL Diffusion kernel}{369}} -\newlabel{ch17:listing:collembola-reduc}{{16.2}{370}} -\@writefile{lol}{\contentsline {lstlisting}{\numberline {16.2}Collembola OpenCL reduction kernel}{370}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.3.3}Collembola performance}{371}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.2}{\ignorespaces Performances CPU et GPU du mod\IeC {\`e}le Collemboles\relax }}{371}} -\newlabel{ch17:fig:mior_perfs_collem}{{16.2}{371}} -\@writefile{toc}{\contentsline {section}{\numberline {16.4}Second example}{372}} -\newlabel{ch17:sec:2ndmodel}{{16.4}{372}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.4.1}The MIOR model}{372}} -\newlabel{ch17:subsec:miormodel}{{16.4.1}{372}} -\@writefile{loa}{\contentsline {algocf}{\numberline {16}{\ignorespaces Evolution step of each Meta-Mior (microbial colony) agent\relax }}{373}} -\newlabel{ch17:seqalgo}{{16}{373}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.4.2}MIOR Implementation}{373}} -\@writefile{toc}{\contentsline {subsubsection}{\numberline {16.4.2.1}Execution mapping on GPU}{373}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.3}{\ignorespaces Execution distribution retained on GPU\relax }}{374}} -\newlabel{ch17:fig:gpu_distribution}{{16.3}{374}} -\@writefile{toc}{\contentsline {subsubsection}{\numberline {16.4.2.2}Data structures translation}{374}} -\newlabel{ch17:subsec:datastructures}{{16.4.2.2}{374}} -\newlabel{ch17:listing:mior_data_structures}{{16.3}{375}} -\@writefile{lol}{\contentsline {lstlisting}{\numberline {16.3}Main data structures used in a MIOR simulation}{375}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.4}{\ignorespaces Compact representation of the topology of a MIOR simulation\relax }}{376}} -\newlabel{ch17:fig:csr_representation}{{16.4}{376}} -\@writefile{toc}{\contentsline {subsubsection}{\numberline {16.4.2.3}Critical resources access management}{376}} -\newlabel{ch17:subsec:concurrency}{{16.4.2.3}{376}} -\newlabel{ch17:listing:mior_kernels}{{16.4}{376}} -\@writefile{lol}{\contentsline {lstlisting}{\numberline {16.4}Main MIOR kernel}{376}} -\newlabel{ch17:fig:mior_launcher}{{16.5}{377}} -\@writefile{lol}{\contentsline {lstlisting}{\numberline {16.5}MIOR simulation launcher}{377}} -\@writefile{toc}{\contentsline {subsubsection}{\numberline {16.4.2.4}Termination detection}{378}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.4.3}Performance of MIOR implementations}{378}} -\newlabel{ch17:subsec:miorexperiments}{{16.4.3}{378}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.5}{\ignorespaces CPU and GPU performance on a Tesla C1060 node\relax }}{380}} -\newlabel{ch17:fig:mior_perfs_tesla}{{16.5}{380}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.6}{\ignorespaces CPU and GPU performance on a personal computer with a Geforce 8800GT\relax }}{381}} -\newlabel{ch17:fig:mior_perfs_8800gt}{{16.6}{381}} -\@writefile{toc}{\contentsline {section}{\numberline {16.5}Analysis and recommandations}{381}} -\newlabel{ch17:analysis}{{16.5}{381}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.5.1}Analysis}{381}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.7}{\ignorespaces Execution time of one multi-simulation kernel on the Tesla platform\relax }}{382}} -\newlabel{ch17:fig:monokernel_graph}{{16.7}{382}} -\@writefile{lof}{\contentsline {figure}{\numberline {16.8}{\ignorespaces Total execution time for 1000 simulations on the Tesla platform, while varying the number of simulations for each kernel\relax }}{382}} -\newlabel{ch17:fig:multikernel_graph}{{16.8}{382}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.5.2}MAS execution workflow}{382}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.5.3}Implementation challenges}{383}} -\@writefile{toc}{\contentsline {subsection}{\numberline {16.5.4}MCSMA}{384}} -\newlabel{ch17:Mcsma}{{16.5.4}{384}} -\@writefile{toc}{\contentsline {section}{\numberline {16.6}Conclusion}{385}} -\newlabel{ch17:conclusion}{{16.6}{385}} -\@writefile{toc}{\contentsline {section}{Bibliography}{386}} +\newlabel{chapter17}{{18}{418}} +\@writefile{toc}{\contentsline {section}{\numberline {18.1}Introduction}{418}} +\newlabel{ch17:intro}{{18.1}{418}} +\@writefile{toc}{\contentsline {section}{\numberline {18.2}Running Agent-Based Simulations}{419}} +\newlabel{ch17:ABM}{{18.2}{419}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.2.1}Multi-agent systems and parallelism}{419}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.2.2}MAS Implementation on GPU}{421}} +\newlabel{ch17:subsec:gpu}{{18.2.2}{421}} +\@writefile{toc}{\contentsline {section}{\numberline {18.3}A first practical example}{422}} +\newlabel{ch17:sec:1stmodel}{{18.3}{422}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.3.1}The Collembola model}{422}} +\newlabel{ch17:subsec:collembolamodel}{{18.3.1}{422}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.1}{\ignorespaces Evolution algorithm of Collembola model\relax }}{423}} +\newlabel{ch17:fig:collem_algorithm}{{18.1}{423}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.3.2}Collembola Implementation}{423}} +\newlabel{ch17:listing:collembola-diffuse}{{18.1}{424}} +\@writefile{lol}{\contentsline {lstlisting}{\numberline {18.1}Collembola OpenCL Diffusion kernel}{424}} +\newlabel{ch17:listing:collembola-reduc}{{18.2}{424}} +\@writefile{lol}{\contentsline {lstlisting}{\numberline {18.2}Collembola OpenCL reduction kernel}{424}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.3.3}Collembola performance}{425}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.2}{\ignorespaces Performance of the Collembola model on CPU and GPU\relax }}{426}} +\newlabel{ch17:fig:mior_perfs_collem}{{18.2}{426}} +\@writefile{toc}{\contentsline {section}{\numberline {18.4}Second example}{426}} +\newlabel{ch17:sec:2ndmodel}{{18.4}{426}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.4.1}The MIOR model}{426}} +\newlabel{ch17:subsec:miormodel}{{18.4.1}{426}} +\@writefile{loa}{\contentsline {algocf}{\numberline {22}{\ignorespaces Evolution step of each Meta-Mior (microbial colony) agent\relax }}{427}} +\newlabel{ch17:seqalgo}{{22}{427}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.4.2}MIOR Implementation}{427}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.3}{\ignorespaces Execution distribution retained on GPU\relax }}{428}} +\newlabel{ch17:fig:gpu_distribution}{{18.3}{428}} +\@writefile{toc}{\contentsline {subsubsection}{\numberline {18.4.2.1}Execution mapping on GPU}{428}} +\@writefile{toc}{\contentsline {subsubsection}{\numberline {18.4.2.2}Data structures translation}{429}} +\newlabel{ch17:subsec:datastructures}{{18.4.2.2}{429}} +\newlabel{ch17:listing:mior_data_structures}{{18.3}{429}} +\@writefile{lol}{\contentsline {lstlisting}{\numberline {18.3}Main data structures used in a MIOR simulation}{429}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.4}{\ignorespaces Compact representation of the topology of a MIOR simulation\relax }}{430}} +\newlabel{ch17:fig:csr_representation}{{18.4}{430}} +\@writefile{toc}{\contentsline {subsubsection}{\numberline {18.4.2.3}Critical resources access management}{430}} +\newlabel{ch17:subsec:concurrency}{{18.4.2.3}{430}} +\newlabel{ch17:listing:mior_kernels}{{18.4}{431}} +\@writefile{lol}{\contentsline {lstlisting}{\numberline {18.4}Main MIOR kernel}{431}} +\newlabel{ch17:fig:mior_launcher}{{18.5}{432}} +\@writefile{lol}{\contentsline {lstlisting}{\numberline {18.5}MIOR simulation launcher}{432}} +\@writefile{toc}{\contentsline {subsubsection}{\numberline {18.4.2.4}Termination detection}{432}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.4.3}Performance of MIOR implementations}{433}} +\newlabel{ch17:subsec:miorexperiments}{{18.4.3}{433}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.5}{\ignorespaces CPU and GPU performance on a Tesla C1060 node\relax }}{434}} +\newlabel{ch17:fig:mior_perfs_tesla}{{18.5}{434}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.6}{\ignorespaces CPU and GPU performance on a personal computer with a Geforce 8800GT\relax }}{435}} +\newlabel{ch17:fig:mior_perfs_8800gt}{{18.6}{435}} +\@writefile{toc}{\contentsline {section}{\numberline {18.5}Analysis and recommendations}{435}} +\newlabel{ch17:analysis}{{18.5}{435}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.7}{\ignorespaces Execution time of one multi-simulation kernel on the Tesla platform\relax }}{436}} +\newlabel{ch17:fig:monokernel_graph}{{18.7}{436}} +\@writefile{lof}{\contentsline {figure}{\numberline {18.8}{\ignorespaces Total execution time for 1000 simulations on the Tesla platform, while varying the number of simulations for each kernel\relax }}{436}} +\newlabel{ch17:fig:multikernel_graph}{{18.8}{436}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.5.1}Analysis}{436}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.5.2}MAS execution workflow}{437}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.5.3}Implementation challenges}{438}} +\@writefile{toc}{\contentsline {subsection}{\numberline {18.5.4}MCSMA}{438}} +\newlabel{ch17:Mcsma}{{18.5.4}{438}} +\@writefile{toc}{\contentsline {section}{\numberline {18.6}Conclusion}{439}} +\newlabel{ch17:conclusion}{{18.6}{439}} +\@writefile{toc}{\contentsline {section}{Bibliography}{440}} \@setckpt{Chapters/chapter17/ch17}{ -\setcounter{page}{389} +\setcounter{page}{444} \setcounter{equation}{0} \setcounter{enumi}{3} \setcounter{enumii}{0} @@ -79,7 +84,7 @@ \setcounter{footnote}{1} \setcounter{mpfootnote}{0} \setcounter{part}{6} -\setcounter{chapter}{16} +\setcounter{chapter}{18} \setcounter{section}{6} \setcounter{subsection}{0} \setcounter{subsubsection}{0} @@ -96,9 +101,9 @@ \setcounter{lstnumber}{21} \setcounter{ContinuedFloat}{0} \setcounter{AlgoLine}{17} -\setcounter{algocfline}{16} -\setcounter{algocfproc}{16} -\setcounter{algocf}{16} +\setcounter{algocfline}{22} +\setcounter{algocfproc}{22} +\setcounter{algocf}{22} \setcounter{nprt@mantissa@digitsbefore}{0} \setcounter{nprt@mantissa@digitsafter}{0} \setcounter{nprt@exponent@digitsbefore}{0}