From: couturie Date: Fri, 29 Apr 2016 13:58:02 +0000 (+0200) Subject: new X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/kahina_paper2.git/commitdiff_plain/9a2e466ad0fe6f247f15f228b5b76f5854ccb0ca new --- diff --git a/maj.tex b/maj.tex index fe899a3..c9e308c 100644 --- a/maj.tex +++ b/maj.tex @@ -32,7 +32,7 @@ Email: \{kahina.ghidouche,ar.sider\}@univ-bejaia.dz} \and \IEEEauthorblockN{Lilia Ziane Khodja, Raphaël Couturier} \IEEEauthorblockA{FEMTO-ST Institute\\ - University of Bourgogne Franche-Comte, France\\ + University Bourgogne Franche-Comte, France\\ Email: zianekhodja.lilia@gmail.com\\ raphael.couturier@univ-fcomte.fr}} @@ -275,10 +275,10 @@ where \verb=DBL_MAX= stands for the maximum representable \subsection{The Ehrlich-Aberth parallel implementation on CUDA} -\KG{Je viens de refaire cette section} The algorithm ~\ref{alg1-cuda} shows sketch of the Ehrlich-Aberth method using CUDA. -The first steps consist in the initialization of the input data like, the polynomial P,derivative of P and the vector solution Z. Then, all data of the root finding problem -must be copied from the CPU memory to the GPU global memory,because +The first step consists in the initialization of the input data, for +exemple the polynomial P, the derivative of P and the vector solution Z. Then, all data of the root finding problem +are transfered from the CPU memory to the GPU global memory, because the GPUs only work on the data filled in their memories. Next, all the data-parallel arithmetic operations inside the main loop \verb=(while(...))= are executed as kernels by the GPU. The @@ -287,7 +287,7 @@ order to check the convergence of the roots after each iteration (line 7, Algorithm~\ref{alg1-cuda}). Then the new roots with the new iterations are computed using the EA method with a Gauss-Seidel iteration mode in order to use the latest updated roots (line -6). This improves the convergence. This kernel is, in practice, very +6). This improves the convergence compared to the Jacobi method. This kernel is, in practice, very long since it performs all the operations with complex numbers with the normal mode of the EA method as in Eq.~\ref{Eq:EA1} but also with the logarithm-exponential one as in Eq.(~\ref{Log_H1},~\ref{Log_H2}). The last kernel checks the convergence of the roots after each update of $Z^{k}$, according to formula Eq.~\ref{eq:Aberth-Conv-Cond} line (7). We used the functions of the CUBLAS Library (CUDA Basic Linear Algebra Subroutines) to implement this kernel. The algorithm terminates its computations when all the roots have converged. @@ -310,7 +310,7 @@ converged. %are transferred from the GPU to the CPU. \begin{algorithm}[htpb] \label{alg1-cuda} -%\LinesNumbered +\LinesNumbered %\SetAlgoNoLine \caption{Finding roots of polynomials with the Ehrlich-Aberth method on a GPU} \KwIn{ $\epsilon$ (tolerance threshold)} @@ -326,7 +326,7 @@ Copy $P$, $P'$ and $Z$ from CPU to GPU\; Copy $Z$ from GPU to CPU\; \end{algorithm} -This figure shows the second kernel code +\RC{A revoir (c'est de la blague en l'état) : Figure~\ref{fig:00} shows the second kernel code} \begin{figure}[htbp] \centering \includegraphics[angle=+0,width=0.4\textwidth]{code1} @@ -347,7 +347,9 @@ Ehrlich-Aberth method with OpenMP and MPI is presented. \section{The Ehrlich-Aberth algorithm on multiple GPUs} \label{sec4} -\KG{we remind that to manage the CUDA contexts of different GPUs, We investigate two parallel paradigms: OpenMP and MPI. In this section we present the both \textit{OpenMP-CUDA} approach and the \textit{MPI-CUDA} approach used to implement the Ehrlich-Aberth algorithm on multiple GPUs.} +In order to manage the CUDA contexts of different GPUs, two parallel +paradigms are investigated: OpenMP and MPI. In this section we present + the \textit{OpenMP-CUDA} and the \textit{MPI-CUDA} approaches used to implement the Ehrlich-Aberth algorithm on multiple GPUs. \subsection{An OpenMP-CUDA approach} Our OpenMP-CUDA implementation of EA algorithm is based on the hybrid @@ -381,7 +383,7 @@ memory arrays containing all the roots. \begin{algorithm}[htpb] \label{alg2-cuda-openmp} -%\LinesNumbered +\LinesNumbered %\SetAlgoNoLine \caption{Finding roots of polynomials with the Ehrlich-Aberth method on multiple GPUs using OpenMP} \KwIn{ $\epsilon$ (tolerance threshold)} @@ -432,7 +434,7 @@ new iteration is executed. \begin{algorithm}[htpb] \label{alg2-cuda-mpi} -%\LinesNumbered +\LinesNumbered %\SetAlgoNoLine \caption{Finding roots of polynomials with the Ehrlich-Aberth method on multiple GPUs using MPI} @@ -471,8 +473,8 @@ We study two categories of polynomials: sparse polynomials and full polynomials. {\Large \forall \alpha_{i} \in \mathbb{C}, i\in \mathbb{N}; p(x)=\sum^{n}_{i=0} \alpha_{i}.x^{i}} \end{equation} -\KG{For our tests, a CPU Intel(R) Xeon(R) CPU -X5650@2.40GHz and 4 GPUs cards Tesla Kepler K40,are used with CUDA version 7.5}. +For our tests, a CPU Intel(R) Xeon(R) CPU +X5650@2.40GHz and 4 GPUs cards Tesla Kepler K40, are used with CUDA version 7.5. In order to evaluate both the GPU and Multi-GPU approaches, we performed a set of experiments on a single GPU and multiple GPUs using OpenMP or MPI with @@ -585,7 +587,7 @@ Our next objective is to extend the model presented here to clusters of GPU node \section*{Acknowledgment} This paper is partially funded by the Labex ACTION program (contract -ANR-11-LABX-01-01). Computations have been performed on the supercomputer facilities of the Mésocentre de calcul de Franche-Comté. We also would like to thank Nvidia for hardware donation under CUDA Research Center 2014. +ANR-11-LABX-01-01) and the Franche-Comte regional council. Computations have been performed on the supercomputer facilities of the Mésocentre de calcul de Franche-Comté. We also would like to thank Nvidia for hardware donation under CUDA Research Center 2014. \bibliography{mybibfile} diff --git a/paper.tex b/paper.tex index a18d53b..5e2021f 100644 --- a/paper.tex +++ b/paper.tex @@ -275,7 +275,8 @@ where \verb=DBL_MAX= stands for the maximum representable \subsection{The Ehrlich-Aberth parallel implementation on CUDA} -\KG{The algorithm ~\ref{alg1-cuda} shows sketch of the Ehrlich-Aberth method using CUDA. +%\KG{ + The algorithm ~\ref{alg1-cuda} shows sketch of the Ehrlich-Aberth method using CUDA. The first steps consist in the initialization of the input data like, the polynomial P,derivative of P and the vector solution Z. Then, all data of the root finding problem must be copied from the CPU memory to the GPU global memory,because the GPUs only work on the data filled in their memories. @@ -291,7 +292,10 @@ long since it performs all the operations with complex numbers with the normal mode of the EA method as in Eq.~\ref{Eq:EA1} but also with the logarithm-exponential one as in Eq.(~\ref{Log_H1},~\ref{Log_H2}). The last kernel checks the convergence of the roots after each update of $Z^{k}$, according to formula Eq.~\ref{eq:Aberth-Conv-Cond} line (7). We used the functions of the CUBLAS Library (CUDA Basic Linear Algebra Subroutines) to implement this kernel. The algorithm terminates its computations when all the roots have -converged.} +converged. +%} + + %The code is organized as kernels which are parts of codes that are run %on GPU devices. Algorithm~\ref{alg1-cuda} describes the CUDA %implementation of the Ehrlich-Aberth on a GPU. This algorithms starts @@ -326,7 +330,7 @@ Copy $P$, $P'$ and $Z$ from CPU to GPU\; } Copy $Z$ from GPU to CPU\; \end{algorithm} -\\ +\ \\ This figure shows the second kernel code \begin{figure}[htbp] \centering