\@writefile{toc}{\author{Rapha\IeC {\"e}l Couturier}{}}
\@writefile{toc}{\author{Jacques Bahi}{}}
\@writefile{loa}{\addvspace {10\p@ }}
-\@writefile{toc}{\contentsline {chapter}{\numberline {11}Solving sparse linear systems with GMRES and CG methods on GPU clusters}{259}}
+\@writefile{toc}{\contentsline {chapter}{\numberline {12}Solving sparse linear systems with GMRES and CG methods on GPU clusters}{291}}
\@writefile{lof}{\addvspace {10\p@ }}
\@writefile{lot}{\addvspace {10\p@ }}
-\newlabel{ch12}{{11}{259}}
-\@writefile{toc}{\contentsline {section}{\numberline {11.1}Introduction}{259}}
-\newlabel{ch12:sec:01}{{11.1}{259}}
-\@writefile{toc}{\contentsline {section}{\numberline {11.2}Krylov iterative methods}{260}}
-\newlabel{ch12:sec:02}{{11.2}{260}}
-\newlabel{ch12:eq:01}{{11.1}{260}}
-\newlabel{ch12:eq:02}{{11.2}{260}}
-\newlabel{ch12:eq:03}{{11.3}{260}}
-\newlabel{ch12:eq:11}{{11.4}{261}}
-\@writefile{toc}{\contentsline {subsection}{\numberline {11.2.1}CG method}{261}}
-\newlabel{ch12:sec:02.01}{{11.2.1}{261}}
-\newlabel{ch12:eq:04}{{11.5}{261}}
-\newlabel{ch12:eq:05}{{11.6}{261}}
-\newlabel{ch12:eq:06}{{11.7}{261}}
-\newlabel{ch12:eq:07}{{11.8}{261}}
-\newlabel{ch12:eq:08}{{11.9}{261}}
-\newlabel{ch12:eq:09}{{11.10}{261}}
-\@writefile{loa}{\contentsline {algocf}{\numberline {9}{\ignorespaces Left-preconditioned CG method\relax }}{262}}
-\newlabel{ch12:alg:01}{{9}{262}}
-\newlabel{ch12:eq:10}{{11.11}{262}}
-\@writefile{toc}{\contentsline {subsection}{\numberline {11.2.2}GMRES method}{263}}
-\newlabel{ch12:sec:02.02}{{11.2.2}{263}}
-\newlabel{ch12:eq:12}{{11.12}{263}}
-\newlabel{ch12:eq:13}{{11.13}{263}}
-\newlabel{ch12:eq:14}{{11.14}{263}}
-\newlabel{ch12:eq:15}{{11.15}{263}}
-\newlabel{ch12:eq:16}{{11.16}{263}}
-\newlabel{ch12:eq:17}{{11.17}{263}}
-\newlabel{ch12:eq:18}{{11.18}{263}}
-\newlabel{ch12:eq:19}{{11.19}{263}}
-\@writefile{loa}{\contentsline {algocf}{\numberline {10}{\ignorespaces Left-preconditioned GMRES method with restarts\relax }}{264}}
-\newlabel{ch12:alg:02}{{10}{264}}
-\@writefile{toc}{\contentsline {section}{\numberline {11.3}Parallel implementation on a GPU cluster}{265}}
-\newlabel{ch12:sec:03}{{11.3}{265}}
-\@writefile{toc}{\contentsline {subsection}{\numberline {11.3.1}Data partitioning}{265}}
-\newlabel{ch12:sec:03.01}{{11.3.1}{265}}
-\@writefile{lof}{\contentsline {figure}{\numberline {11.1}{\ignorespaces A data partitioning of the sparse matrix $A$, the solution vector $x$ and the right-hand side $b$ into four portions.\relax }}{266}}
-\newlabel{ch12:fig:01}{{11.1}{266}}
-\@writefile{toc}{\contentsline {subsection}{\numberline {11.3.2}GPU computing}{266}}
-\newlabel{ch12:sec:03.02}{{11.3.2}{266}}
-\@writefile{toc}{\contentsline {subsection}{\numberline {11.3.3}Data communications}{267}}
-\newlabel{ch12:sec:03.03}{{11.3.3}{267}}
-\@writefile{lof}{\contentsline {figure}{\numberline {11.2}{\ignorespaces Data exchanges between \textit {Node 1} and its neighbors \textit {Node 0}, \textit {Node 2} and \textit {Node 3}.\relax }}{268}}
-\newlabel{ch12:fig:02}{{11.2}{268}}
-\@writefile{lof}{\contentsline {figure}{\numberline {11.3}{\ignorespaces Columns reordering of a sparse sub-matrix.\relax }}{269}}
-\newlabel{ch12:fig:03}{{11.3}{269}}
-\@writefile{toc}{\contentsline {section}{\numberline {11.4}Experimental results}{270}}
-\newlabel{ch12:sec:04}{{11.4}{270}}
-\@writefile{lof}{\contentsline {figure}{\numberline {11.4}{\ignorespaces General scheme of the GPU cluster of tests composed of six machines, each with two GPUs.\relax }}{270}}
-\newlabel{ch12:fig:04}{{11.4}{270}}
-\@writefile{lof}{\contentsline {figure}{\numberline {11.5}{\ignorespaces Sketches of sparse matrices chosen from the Davis collection.\relax }}{271}}
-\newlabel{ch12:fig:05}{{11.5}{271}}
-\@writefile{lot}{\contentsline {table}{\numberline {11.1}{\ignorespaces Main characteristics of sparse matrices chosen from the Davis collection.\relax }}{271}}
-\newlabel{ch12:tab:01}{{11.1}{271}}
-\@writefile{lot}{\contentsline {table}{\numberline {11.2}{\ignorespaces Performances of the parallel CG method on a cluster of 24 CPU cores vs. on a cluster of 12 GPUs.\relax }}{272}}
-\newlabel{ch12:tab:02}{{11.2}{272}}
-\@writefile{lot}{\contentsline {table}{\numberline {11.3}{\ignorespaces Performances of the parallel GMRES method on a cluster 24 CPU cores vs. on cluster of 12 GPUs.\relax }}{272}}
-\newlabel{ch12:tab:03}{{11.3}{272}}
-\newlabel{ch12:eq:20}{{11.20}{273}}
-\@writefile{lof}{\contentsline {figure}{\numberline {11.6}{\ignorespaces Parallel generation of a large sparse matrix by four computing nodes.\relax }}{274}}
-\newlabel{ch12:fig:06}{{11.6}{274}}
-\@writefile{lot}{\contentsline {table}{\numberline {11.4}{\ignorespaces Main characteristics of sparse banded matrices generated from those of the Davis collection.\relax }}{274}}
-\newlabel{ch12:tab:04}{{11.4}{274}}
-\@writefile{lot}{\contentsline {table}{\numberline {11.5}{\ignorespaces Performances of the parallel CG method for solving linear systems associated to sparse banded matrices on a cluster of 24 CPU cores vs. on a cluster of 12 GPUs.\relax }}{275}}
-\newlabel{ch12:tab:05}{{11.5}{275}}
-\@writefile{toc}{\contentsline {section}{\numberline {11.5}Conclusion}{275}}
-\newlabel{ch12:sec:05}{{11.5}{275}}
-\@writefile{lot}{\contentsline {table}{\numberline {11.6}{\ignorespaces Performances of the parallel GMRES method for solving linear systems associated to sparse banded matrices on a cluster of 24 CPU cores vs. on a cluster of 12 GPUs.\relax }}{276}}
-\newlabel{ch12:tab:06}{{11.6}{276}}
-\@writefile{toc}{\contentsline {section}{Bibliography}{276}}
+\newlabel{ch12}{{12}{291}}
+\@writefile{toc}{\contentsline {section}{\numberline {12.1}Introduction}{291}}
+\newlabel{ch12:sec:01}{{12.1}{291}}
+\@writefile{toc}{\contentsline {section}{\numberline {12.2}Krylov iterative methods}{292}}
+\newlabel{ch12:sec:02}{{12.2}{292}}
+\newlabel{ch12:eq:01}{{12.1}{292}}
+\newlabel{ch12:eq:02}{{12.2}{292}}
+\newlabel{ch12:eq:03}{{12.3}{292}}
+\newlabel{ch12:eq:11}{{12.4}{293}}
+\@writefile{toc}{\contentsline {subsection}{\numberline {12.2.1}CG method}{293}}
+\newlabel{ch12:sec:02.01}{{12.2.1}{293}}
+\newlabel{ch12:eq:04}{{12.5}{293}}
+\newlabel{ch12:eq:05}{{12.6}{293}}
+\newlabel{ch12:eq:06}{{12.7}{293}}
+\newlabel{ch12:eq:07}{{12.8}{293}}
+\newlabel{ch12:eq:08}{{12.9}{293}}
+\newlabel{ch12:eq:09}{{12.10}{293}}
+\@writefile{loa}{\contentsline {algocf}{\numberline {12}{\ignorespaces Left-preconditioned CG method\relax }}{294}}
+\newlabel{ch12:alg:01}{{12}{294}}
+\newlabel{ch12:eq:10}{{12.11}{294}}
+\@writefile{toc}{\contentsline {subsection}{\numberline {12.2.2}GMRES method}{295}}
+\newlabel{ch12:sec:02.02}{{12.2.2}{295}}
+\newlabel{ch12:eq:12}{{12.12}{295}}
+\newlabel{ch12:eq:13}{{12.13}{295}}
+\newlabel{ch12:eq:14}{{12.14}{295}}
+\newlabel{ch12:eq:15}{{12.15}{295}}
+\newlabel{ch12:eq:16}{{12.16}{295}}
+\newlabel{ch12:eq:17}{{12.17}{295}}
+\newlabel{ch12:eq:18}{{12.18}{295}}
+\newlabel{ch12:eq:19}{{12.19}{295}}
+\@writefile{loa}{\contentsline {algocf}{\numberline {13}{\ignorespaces Left-preconditioned GMRES method with restarts\relax }}{296}}
+\newlabel{ch12:alg:02}{{13}{296}}
+\@writefile{toc}{\contentsline {section}{\numberline {12.3}Parallel implementation on a GPU cluster}{297}}
+\newlabel{ch12:sec:03}{{12.3}{297}}
+\@writefile{toc}{\contentsline {subsection}{\numberline {12.3.1}Data partitioning}{297}}
+\newlabel{ch12:sec:03.01}{{12.3.1}{297}}
+\@writefile{lof}{\contentsline {figure}{\numberline {12.1}{\ignorespaces A data partitioning of the sparse matrix $A$, the solution vector $x$ and the right-hand side $b$ into four portions.\relax }}{298}}
+\newlabel{ch12:fig:01}{{12.1}{298}}
+\@writefile{toc}{\contentsline {subsection}{\numberline {12.3.2}GPU computing}{298}}
+\newlabel{ch12:sec:03.02}{{12.3.2}{298}}
+\@writefile{toc}{\contentsline {subsection}{\numberline {12.3.3}Data communications}{299}}
+\newlabel{ch12:sec:03.03}{{12.3.3}{299}}
+\@writefile{lof}{\contentsline {figure}{\numberline {12.2}{\ignorespaces Data exchanges between \textit {Node 1} and its neighbors \textit {Node 0}, \textit {Node 2} and \textit {Node 3}.\relax }}{300}}
+\newlabel{ch12:fig:02}{{12.2}{300}}
+\@writefile{lof}{\contentsline {figure}{\numberline {12.3}{\ignorespaces Columns reordering of a sparse sub-matrix.\relax }}{301}}
+\newlabel{ch12:fig:03}{{12.3}{301}}
+\@writefile{toc}{\contentsline {section}{\numberline {12.4}Experimental results}{302}}
+\newlabel{ch12:sec:04}{{12.4}{302}}
+\@writefile{lof}{\contentsline {figure}{\numberline {12.4}{\ignorespaces General scheme of the GPU cluster of tests composed of six machines, each with two GPUs.\relax }}{302}}
+\newlabel{ch12:fig:04}{{12.4}{302}}
+\@writefile{lof}{\contentsline {figure}{\numberline {12.5}{\ignorespaces Sketches of sparse matrices chosen from the Davis collection.\relax }}{303}}
+\newlabel{ch12:fig:05}{{12.5}{303}}
+\@writefile{lot}{\contentsline {table}{\numberline {12.1}{\ignorespaces Main characteristics of sparse matrices chosen from the Davis collection.\relax }}{303}}
+\newlabel{ch12:tab:01}{{12.1}{303}}
+\@writefile{lot}{\contentsline {table}{\numberline {12.2}{\ignorespaces Performances of the parallel CG method on a cluster of 24 CPU cores vs. on a cluster of 12 GPUs.\relax }}{304}}
+\newlabel{ch12:tab:02}{{12.2}{304}}
+\@writefile{lot}{\contentsline {table}{\numberline {12.3}{\ignorespaces Performances of the parallel GMRES method on a cluster 24 CPU cores vs. on cluster of 12 GPUs.\relax }}{304}}
+\newlabel{ch12:tab:03}{{12.3}{304}}
+\newlabel{ch12:eq:20}{{12.20}{305}}
+\@writefile{lof}{\contentsline {figure}{\numberline {12.6}{\ignorespaces Parallel generation of a large sparse matrix by four computing nodes.\relax }}{306}}
+\newlabel{ch12:fig:06}{{12.6}{306}}
+\@writefile{lot}{\contentsline {table}{\numberline {12.4}{\ignorespaces Main characteristics of sparse banded matrices generated from those of the Davis collection.\relax }}{306}}
+\newlabel{ch12:tab:04}{{12.4}{306}}
+\@writefile{lot}{\contentsline {table}{\numberline {12.5}{\ignorespaces Performances of the parallel CG method for solving linear systems associated to sparse banded matrices on a cluster of 24 CPU cores vs. on a cluster of 12 GPUs.\relax }}{307}}
+\newlabel{ch12:tab:05}{{12.5}{307}}
+\@writefile{toc}{\contentsline {section}{\numberline {12.5}Conclusion}{307}}
+\newlabel{ch12:sec:05}{{12.5}{307}}
+\@writefile{lot}{\contentsline {table}{\numberline {12.6}{\ignorespaces Performances of the parallel GMRES method for solving linear systems associated to sparse banded matrices on a cluster of 24 CPU cores vs. on a cluster of 12 GPUs.\relax }}{308}}
+\newlabel{ch12:tab:06}{{12.6}{308}}
+\@writefile{toc}{\contentsline {section}{Bibliography}{308}}
\@setckpt{Chapters/chapter12/ch12}{
-\setcounter{page}{278}
+\setcounter{page}{310}
\setcounter{equation}{22}
-\setcounter{enumi}{4}
+\setcounter{enumi}{2}
\setcounter{enumii}{0}
\setcounter{enumiii}{0}
\setcounter{enumiv}{10}
\setcounter{footnote}{0}
\setcounter{mpfootnote}{0}
\setcounter{part}{5}
-\setcounter{chapter}{11}
+\setcounter{chapter}{12}
\setcounter{section}{5}
\setcounter{subsection}{0}
\setcounter{subsubsection}{0}
\setcounter{lstnumber}{50}
\setcounter{ContinuedFloat}{0}
\setcounter{AlgoLine}{29}
-\setcounter{algocfline}{10}
-\setcounter{algocfproc}{10}
-\setcounter{algocf}{10}
+\setcounter{algocfline}{13}
+\setcounter{algocfproc}{13}
+\setcounter{algocf}{13}
\setcounter{nprt@mantissa@digitsbefore}{0}
\setcounter{nprt@mantissa@digitsafter}{0}
\setcounter{nprt@exponent@digitsbefore}{0}
