The Experts below are selected from a list of 138870 Experts worldwide ranked by ideXlab platform
David E Keyes - One of the best experts on this subject based on the ideXlab platform.
-
asynchronous task based polar decomposition on single node manycore architectures
IEEE Transactions on Parallel and Distributed Systems, 2018Co-Authors: Dalal Sukkari, Hatem Ltaief, Mathieu Faverge, David E KeyesAbstract:This paper introduces the first asynchronous, task-based formulation of the polar decomposition and its corresponding implementation on manycore architectures. Based on a new formulation of the iterative $QR$ dynamically-weighted Halley algorithm (QDWH) for the calculation of the polar decomposition, the proposed implementation replaces the original and hostile LU factorization for the condition number estimator by the more adequate $QR$ factorization to enable software portability across various architectures. Relying on fine-grained computations, the novel task-based implementation is also capable of taking advantage of the Identity Structure of the matrix involved during the QDWH iterations, which decreases the overall algorithmic complexity. Furthermore, the artifactual synchronization points have been weakened compared to previous implementations, unveiling look-ahead opportunities for better hardware occupancy. The overall QDWH-based polar decomposition can then be represented as a directed acyclic graph (DAG), where nodes represent computational tasks and edges define the inter-task data dependencies. The StarPU dynamic runtime system is employed to traverse the DAG, to track the various data dependencies and to asynchronously schedule the computational tasks on the underlying hardware resources, resulting in an out-of-order task scheduling. Benchmarking experiments show significant improvements against existing state-of-the-art high performance implementations (i.e., Intel MKL and Elemental) for the polar decomposition on latest shared-memory vendors’ systems (i.e., Intel Haswell/Broadwell/Knights Landing, NVIDIA K80/P100 GPUs and IBM Power8), while maintaining high numerical accuracy.
Dalal Sukkari - One of the best experts on this subject based on the ideXlab platform.
-
asynchronous task based polar decomposition on single node manycore architectures
IEEE Transactions on Parallel and Distributed Systems, 2018Co-Authors: Dalal Sukkari, Hatem Ltaief, Mathieu Faverge, David E KeyesAbstract:This paper introduces the first asynchronous, task-based formulation of the polar decomposition and its corresponding implementation on manycore architectures. Based on a new formulation of the iterative $QR$ dynamically-weighted Halley algorithm (QDWH) for the calculation of the polar decomposition, the proposed implementation replaces the original and hostile LU factorization for the condition number estimator by the more adequate $QR$ factorization to enable software portability across various architectures. Relying on fine-grained computations, the novel task-based implementation is also capable of taking advantage of the Identity Structure of the matrix involved during the QDWH iterations, which decreases the overall algorithmic complexity. Furthermore, the artifactual synchronization points have been weakened compared to previous implementations, unveiling look-ahead opportunities for better hardware occupancy. The overall QDWH-based polar decomposition can then be represented as a directed acyclic graph (DAG), where nodes represent computational tasks and edges define the inter-task data dependencies. The StarPU dynamic runtime system is employed to traverse the DAG, to track the various data dependencies and to asynchronously schedule the computational tasks on the underlying hardware resources, resulting in an out-of-order task scheduling. Benchmarking experiments show significant improvements against existing state-of-the-art high performance implementations (i.e., Intel MKL and Elemental) for the polar decomposition on latest shared-memory vendors’ systems (i.e., Intel Haswell/Broadwell/Knights Landing, NVIDIA K80/P100 GPUs and IBM Power8), while maintaining high numerical accuracy.
-
Asynchronous Task-Based Polar Decomposition on Manycore Architectures
2016Co-Authors: Dalal Sukkari, Hatem Ltaief, Mathieu Faverge, David KeyesAbstract:This paper introduces the first asynchronous, task-based implementation of the polar decomposition on manycore architectures. Based on a new formulation of the iterative QR dynamically-weighted Halley algorithm (QDWH) for the calculation of the polar decomposition, the proposed implementation replaces the original and hostile LU factorization for the condition number estimator by the more adequate QR factorization to enable software portability across various architectures. Relying on fine-grained computations, the novel task-based implementation is also capable of taking advantage of the Identity Structure of the matrix involved during the QDWH iterations, which decreases the overall algorithmic complexity. Furthermore, the artifactual synchronization points have been severely weakened compared to previous implementations, unveiling look-ahead opportunities for better hardware occupancy. The overall QDWH-based polar decomposition can then be represented as a directed acyclic graph (DAG), where nodes represent computational tasks and edges define the inter-task data dependencies. The StarPU dynamic runtime system is employed to traverse the DAG, to track the various data dependencies and to asynchronously schedule the computational tasks on the underlying hardware resources, resulting in an out-of-order task scheduling. Benchmarking experiments show significant improvements against existing state-of-the-art high performance implementations (i.e., Intel MKL and Elemental) for the polar decomposition on latest shared-memory vendors' systems (i.e., Intel Haswell/Broadwell/Knights Landing, NVIDIA K80/P100 GPUs and IBM Power8), while maintaining high numerical accuracy.
Hatem Ltaief - One of the best experts on this subject based on the ideXlab platform.
-
asynchronous task based polar decomposition on single node manycore architectures
IEEE Transactions on Parallel and Distributed Systems, 2018Co-Authors: Dalal Sukkari, Hatem Ltaief, Mathieu Faverge, David E KeyesAbstract:This paper introduces the first asynchronous, task-based formulation of the polar decomposition and its corresponding implementation on manycore architectures. Based on a new formulation of the iterative $QR$ dynamically-weighted Halley algorithm (QDWH) for the calculation of the polar decomposition, the proposed implementation replaces the original and hostile LU factorization for the condition number estimator by the more adequate $QR$ factorization to enable software portability across various architectures. Relying on fine-grained computations, the novel task-based implementation is also capable of taking advantage of the Identity Structure of the matrix involved during the QDWH iterations, which decreases the overall algorithmic complexity. Furthermore, the artifactual synchronization points have been weakened compared to previous implementations, unveiling look-ahead opportunities for better hardware occupancy. The overall QDWH-based polar decomposition can then be represented as a directed acyclic graph (DAG), where nodes represent computational tasks and edges define the inter-task data dependencies. The StarPU dynamic runtime system is employed to traverse the DAG, to track the various data dependencies and to asynchronously schedule the computational tasks on the underlying hardware resources, resulting in an out-of-order task scheduling. Benchmarking experiments show significant improvements against existing state-of-the-art high performance implementations (i.e., Intel MKL and Elemental) for the polar decomposition on latest shared-memory vendors’ systems (i.e., Intel Haswell/Broadwell/Knights Landing, NVIDIA K80/P100 GPUs and IBM Power8), while maintaining high numerical accuracy.
-
Asynchronous Task-Based Polar Decomposition on Manycore Architectures
2016Co-Authors: Dalal Sukkari, Hatem Ltaief, Mathieu Faverge, David KeyesAbstract:This paper introduces the first asynchronous, task-based implementation of the polar decomposition on manycore architectures. Based on a new formulation of the iterative QR dynamically-weighted Halley algorithm (QDWH) for the calculation of the polar decomposition, the proposed implementation replaces the original and hostile LU factorization for the condition number estimator by the more adequate QR factorization to enable software portability across various architectures. Relying on fine-grained computations, the novel task-based implementation is also capable of taking advantage of the Identity Structure of the matrix involved during the QDWH iterations, which decreases the overall algorithmic complexity. Furthermore, the artifactual synchronization points have been severely weakened compared to previous implementations, unveiling look-ahead opportunities for better hardware occupancy. The overall QDWH-based polar decomposition can then be represented as a directed acyclic graph (DAG), where nodes represent computational tasks and edges define the inter-task data dependencies. The StarPU dynamic runtime system is employed to traverse the DAG, to track the various data dependencies and to asynchronously schedule the computational tasks on the underlying hardware resources, resulting in an out-of-order task scheduling. Benchmarking experiments show significant improvements against existing state-of-the-art high performance implementations (i.e., Intel MKL and Elemental) for the polar decomposition on latest shared-memory vendors' systems (i.e., Intel Haswell/Broadwell/Knights Landing, NVIDIA K80/P100 GPUs and IBM Power8), while maintaining high numerical accuracy.
Mathieu Faverge - One of the best experts on this subject based on the ideXlab platform.
-
asynchronous task based polar decomposition on single node manycore architectures
IEEE Transactions on Parallel and Distributed Systems, 2018Co-Authors: Dalal Sukkari, Hatem Ltaief, Mathieu Faverge, David E KeyesAbstract:This paper introduces the first asynchronous, task-based formulation of the polar decomposition and its corresponding implementation on manycore architectures. Based on a new formulation of the iterative $QR$ dynamically-weighted Halley algorithm (QDWH) for the calculation of the polar decomposition, the proposed implementation replaces the original and hostile LU factorization for the condition number estimator by the more adequate $QR$ factorization to enable software portability across various architectures. Relying on fine-grained computations, the novel task-based implementation is also capable of taking advantage of the Identity Structure of the matrix involved during the QDWH iterations, which decreases the overall algorithmic complexity. Furthermore, the artifactual synchronization points have been weakened compared to previous implementations, unveiling look-ahead opportunities for better hardware occupancy. The overall QDWH-based polar decomposition can then be represented as a directed acyclic graph (DAG), where nodes represent computational tasks and edges define the inter-task data dependencies. The StarPU dynamic runtime system is employed to traverse the DAG, to track the various data dependencies and to asynchronously schedule the computational tasks on the underlying hardware resources, resulting in an out-of-order task scheduling. Benchmarking experiments show significant improvements against existing state-of-the-art high performance implementations (i.e., Intel MKL and Elemental) for the polar decomposition on latest shared-memory vendors’ systems (i.e., Intel Haswell/Broadwell/Knights Landing, NVIDIA K80/P100 GPUs and IBM Power8), while maintaining high numerical accuracy.
-
Asynchronous Task-Based Polar Decomposition on Manycore Architectures
2016Co-Authors: Dalal Sukkari, Hatem Ltaief, Mathieu Faverge, David KeyesAbstract:This paper introduces the first asynchronous, task-based implementation of the polar decomposition on manycore architectures. Based on a new formulation of the iterative QR dynamically-weighted Halley algorithm (QDWH) for the calculation of the polar decomposition, the proposed implementation replaces the original and hostile LU factorization for the condition number estimator by the more adequate QR factorization to enable software portability across various architectures. Relying on fine-grained computations, the novel task-based implementation is also capable of taking advantage of the Identity Structure of the matrix involved during the QDWH iterations, which decreases the overall algorithmic complexity. Furthermore, the artifactual synchronization points have been severely weakened compared to previous implementations, unveiling look-ahead opportunities for better hardware occupancy. The overall QDWH-based polar decomposition can then be represented as a directed acyclic graph (DAG), where nodes represent computational tasks and edges define the inter-task data dependencies. The StarPU dynamic runtime system is employed to traverse the DAG, to track the various data dependencies and to asynchronously schedule the computational tasks on the underlying hardware resources, resulting in an out-of-order task scheduling. Benchmarking experiments show significant improvements against existing state-of-the-art high performance implementations (i.e., Intel MKL and Elemental) for the polar decomposition on latest shared-memory vendors' systems (i.e., Intel Haswell/Broadwell/Knights Landing, NVIDIA K80/P100 GPUs and IBM Power8), while maintaining high numerical accuracy.
Marilynn B. Brewer - One of the best experts on this subject based on the ideXlab platform.
-
construing multiple ingroups assessing social Identity inclusiveness and Structure in ethnic and religious minority group members
European Journal of Social Psychology, 2015Co-Authors: Andrea Van Dommelen, Katharina Schmid, Miles Hewstone, Karen Gonsalkorale, Marilynn B. BrewerAbstract:The combination of multiple social identities into a coherent ingroup construal is of immediate relevance in today's complex and diverse societies. This paper proposes a conceptual and operational framework to examine how individuals subjectively construe their ingroup in the context of multiple, cross-cutting group memberships. The subjective combination of multiple social identities is described in terms of Structure (social Identity Structure) and inclusiveness (social Identity inclusiveness (SII)). Two studies assess SII and social Identity Structure in community samples to whom the subjective combination of multiple, cross-cutting ingroups is of particular relevance: a sample of Turkish-Belgian Muslims (Study 1) and Turkish-Australian Muslims (Study 2). Across both studies, SII uniquely predicted attitudes toward a range of outgroups, over and above identification with singular ingroups. Moreover, a wide range of social Identity Structures were identified, further attesting to broad individual differences in the construal of the perceived ingroup. Copyright © 2015 John Wiley & Sons, Ltd.
-
Construing multiple in-groups: Assessing social Identity inclusiveness and Structure in ethnic and religious minority group members
European Journal of Social Psychology, 2015Co-Authors: Andrea Van Dommelen, Katharina Schmid, Miles Hewstone, Karen Gonsalkorale, Marilynn B. BrewerAbstract:The combination of multiple social identities into a coherent ingroup construal is of immediate relevance in today's complex and diverse societies. This paper proposes a conceptual and operational framework to examine how individuals subjectively construe their ingroup in the context of multiple, cross-cutting group memberships. The subjective combination of multiple social identities is described in terms of Structure (social Identity Structure) and inclusiveness (social Identity inclusiveness (SII)). Two studies assess SII and social Identity Structure in community samples to whom the subjective combination of multiple, cross-cutting ingroups is of particular relevance: a sample of Turkish-Belgian Muslims (Study 1) and Turkish-Australian Muslims (Study 2). Across both studies, SII uniquely predicted attitudes toward a range of outgroups, over and above identification with singular ingroups. Moreover, a wide range of social Identity Structures were identified, further attesting to broad individual differences in the construal of the perceived ingroup. Copyright © 2015 John Wiley & Sons, Ltd.
-
Social Identity Complexity
Personality and Social Psychology Review, 2002Co-Authors: Sonia Roccas, Marilynn B. BrewerAbstract:In this article, we introduce the concept of social Identity complexity—a new theoretical construct that refers to an individual’s subjective representation of the interrelationships among his or her multiple group identities. Social Identity complexity reflects the degree of overlap perceived to exist between groups of which a person is simultaneously a member. When the overlap of multiple ingroups is perceived to be high, the individual maintains a relatively simplified Identity Structure whereby memberships in different groups converge to form a single ingroup identification. When a person acknowledges, and accepts, that memberships in multiple ingroups are not fully convergent or overlapping, the associated Identity Structure is both more inclusive and more complex. In this article, we define the concept of social Identity complexity and discuss its possible antecedents and consequences. Results from initial studies support the prediction that social Identity complexity is affected by stress and is related to personal value priorities and to tolerance of outgroup members. Recently researchers of group processes have expressed increasing interest in the fact that most individuals are simultaneously members of multiple social groups. Although there has been some research on the
