The Experts below are selected from a list of 1503 Experts worldwide ranked by ideXlab platform
Vincent Pichon - One of the best experts on this subject based on the ideXlab platform.
-
Computing manuscript No. (will be inserted by the editor) A Low Level Component Model Easing Performance Portability of HPC Applications
2014Co-Authors: Julien Bigot, Zhengxiong Hou, Vincent PichonAbstract:Abstract Scientic applications are getting increasingly complex, e.g., to im-prove their accuracy by taking into account more phenomena. Meanwhile, computing infrastructures are continuing their fast evolution. Thus, software engineering is becoming a major issue to offer ease of development, portability and maintainability while achieving high performance. Component based soft-ware engineering offers a promising approach that enables the manipulation of the software architecture of applications. However, existing models do not provide an adequate support for performance portability of HPC applications. This paper proposes a Low Level Component model (L2C) that supports inter-component interactions for typical scenarios of high performance com-puting, such as process-local shared memory and Function Invocation (C++ and Fortran), MPI, and Corba. To study the benets of using L2C, this paper walks through an example of stencil computation, i.e. a structured mesh Jacobi implementation of the 2D heat equation parallelized through domain decomposition. The experimental results obtained on the Grid'5000 testbed and on the Curie supercomputer show that L2C can achieve performance sim-ilar to that of native implementations, while easing performance portability.
-
A low level component model easing performance portability of HPC applications
Computing, 2014Co-Authors: Julien Bigot, Zhengxiong Hou, Christian Perez, Vincent PichonAbstract:Scientific applications are getting increasingly complex, e.g., to improve their accuracy by taking into account more phenomena. Meanwhile, computing infrastructures are continuing their fast evolution. Thus, software engineering is becoming a major issue to offer ease of development, portability and maintainability while achieving high performance. Component based software engineering offers a promising approach that enables the manipulation of the software architecture of applications. However, existing models do not provide an adequate support for performance portability of HPC applications. This paper proposes a low level component model (L $$^2$$ 2 C) that supports inter-component interactions for typical scenarios of high performance computing, such as process-local shared memory and Function Invocation (C++ and Fortran ), MPI , and Corba . To study the benefits of using L $$^2$$ 2 C, this paper walks through an example of stencil computation, i.e. a structured mesh Jacobi implementation of the 2D heat equation parallelized through domain decomposition. The experimental results obtained on the Grid’5000 testbed and on the Curie supercomputer show that L $$^2$$ 2 C can achieve performance similar to that of native implementations, while easing performance portability.
Julien Bigot - One of the best experts on this subject based on the ideXlab platform.
-
Computing manuscript No. (will be inserted by the editor) A Low Level Component Model Easing Performance Portability of HPC Applications
2014Co-Authors: Julien Bigot, Zhengxiong Hou, Vincent PichonAbstract:Abstract Scientic applications are getting increasingly complex, e.g., to im-prove their accuracy by taking into account more phenomena. Meanwhile, computing infrastructures are continuing their fast evolution. Thus, software engineering is becoming a major issue to offer ease of development, portability and maintainability while achieving high performance. Component based soft-ware engineering offers a promising approach that enables the manipulation of the software architecture of applications. However, existing models do not provide an adequate support for performance portability of HPC applications. This paper proposes a Low Level Component model (L2C) that supports inter-component interactions for typical scenarios of high performance com-puting, such as process-local shared memory and Function Invocation (C++ and Fortran), MPI, and Corba. To study the benets of using L2C, this paper walks through an example of stencil computation, i.e. a structured mesh Jacobi implementation of the 2D heat equation parallelized through domain decomposition. The experimental results obtained on the Grid'5000 testbed and on the Curie supercomputer show that L2C can achieve performance sim-ilar to that of native implementations, while easing performance portability.
-
A low level component model easing performance portability of HPC applications
Computing, 2014Co-Authors: Julien Bigot, Zhengxiong Hou, Christian Perez, Vincent PichonAbstract:Scientific applications are getting increasingly complex, e.g., to improve their accuracy by taking into account more phenomena. Meanwhile, computing infrastructures are continuing their fast evolution. Thus, software engineering is becoming a major issue to offer ease of development, portability and maintainability while achieving high performance. Component based software engineering offers a promising approach that enables the manipulation of the software architecture of applications. However, existing models do not provide an adequate support for performance portability of HPC applications. This paper proposes a low level component model (L $$^2$$ 2 C) that supports inter-component interactions for typical scenarios of high performance computing, such as process-local shared memory and Function Invocation (C++ and Fortran ), MPI , and Corba . To study the benefits of using L $$^2$$ 2 C, this paper walks through an example of stencil computation, i.e. a structured mesh Jacobi implementation of the 2D heat equation parallelized through domain decomposition. The experimental results obtained on the Grid’5000 testbed and on the Curie supercomputer show that L $$^2$$ 2 C can achieve performance similar to that of native implementations, while easing performance portability.
Pichon Vincent - One of the best experts on this subject based on the ideXlab platform.
-
A low level component model easing performance portability of HPC applications
'Springer Science and Business Media LLC', 2014Co-Authors: Bigot Julien, Hou Zhengxiong, Pérez Christian, Pichon VincentAbstract:International audienceScientific applications are getting increasingly complex, e.g., to improve their accuracy by taking into account more phenomena. Meanwhile, computing infrastructures are continuing their fast evolution. Thus, software engineering is becoming a major issue to offer ease of development, portability and maintainability while achieving high performance. Component based software engineering offers a promising approach that enables the manipulation of the software architecture of applications. However, existing models do not provide an adequate support for performance portability of HPC applications. This paper proposes a low level component model (L²C) that supports inter-component interactions for typical scenarios of high performance computing, such as process-local shared memory and Function Invocation (C++ and Fortran), MPI, and Corba. To study the benefits of using L²C, this paper walks through an example of stencil computation, i.e. a structured mesh Jacobi implementation of the 2D heat equation parallelized through domain decomposition. The experimental results obtained on the Grid'5000 testbed and on the Curie supercomputer show that L²C can achieve performance similar to that of native implementations, while easing performance portability
Zhengxiong Hou - One of the best experts on this subject based on the ideXlab platform.
-
Computing manuscript No. (will be inserted by the editor) A Low Level Component Model Easing Performance Portability of HPC Applications
2014Co-Authors: Julien Bigot, Zhengxiong Hou, Vincent PichonAbstract:Abstract Scientic applications are getting increasingly complex, e.g., to im-prove their accuracy by taking into account more phenomena. Meanwhile, computing infrastructures are continuing their fast evolution. Thus, software engineering is becoming a major issue to offer ease of development, portability and maintainability while achieving high performance. Component based soft-ware engineering offers a promising approach that enables the manipulation of the software architecture of applications. However, existing models do not provide an adequate support for performance portability of HPC applications. This paper proposes a Low Level Component model (L2C) that supports inter-component interactions for typical scenarios of high performance com-puting, such as process-local shared memory and Function Invocation (C++ and Fortran), MPI, and Corba. To study the benets of using L2C, this paper walks through an example of stencil computation, i.e. a structured mesh Jacobi implementation of the 2D heat equation parallelized through domain decomposition. The experimental results obtained on the Grid'5000 testbed and on the Curie supercomputer show that L2C can achieve performance sim-ilar to that of native implementations, while easing performance portability.
-
A low level component model easing performance portability of HPC applications
Computing, 2014Co-Authors: Julien Bigot, Zhengxiong Hou, Christian Perez, Vincent PichonAbstract:Scientific applications are getting increasingly complex, e.g., to improve their accuracy by taking into account more phenomena. Meanwhile, computing infrastructures are continuing their fast evolution. Thus, software engineering is becoming a major issue to offer ease of development, portability and maintainability while achieving high performance. Component based software engineering offers a promising approach that enables the manipulation of the software architecture of applications. However, existing models do not provide an adequate support for performance portability of HPC applications. This paper proposes a low level component model (L $$^2$$ 2 C) that supports inter-component interactions for typical scenarios of high performance computing, such as process-local shared memory and Function Invocation (C++ and Fortran ), MPI , and Corba . To study the benefits of using L $$^2$$ 2 C, this paper walks through an example of stencil computation, i.e. a structured mesh Jacobi implementation of the 2D heat equation parallelized through domain decomposition. The experimental results obtained on the Grid’5000 testbed and on the Curie supercomputer show that L $$^2$$ 2 C can achieve performance similar to that of native implementations, while easing performance portability.
Andrew A. Chien - One of the best experts on this subject based on the ideXlab platform.
-
precise concrete type inference for object oriented languages
Conference on Object-Oriented Programming Systems Languages and Applications, 1994Co-Authors: John Plevyak, Andrew A. ChienAbstract:Concrete type information is invaluable for program optimization. The determination of concrete types in object-oriented languages is a flow sensitive global data flow problem. It is made difficult by dynamic dispatch (virtual Function Invocation) and first class Functions (and selectors)—the very program structures for whose optimization its results are most critical. Previous work has shown that constraint-based type inference systems can be used to safely approximate concrete types [15], but their use can be expensive and their results imprecise. We present an incremental constraint-based type inference which produces precise concrete type information for a much larger class of programs at lower cost. Our algorithm extends the analysis in response to discovered imprecisions, guiding the analysis' effort to where it is most productive. This produces precise information at a cost proportional to the type complexity of the program. Many programs untypable by previous approaches or practically untypable due to computational expense, can be precisely analyzed by our new algorithm. Performance results, precision, and running time, are reported for a number of concurrent object-oriented programs. These results confirm the algorithm's precision and efficiency.