Parallel input/output impact on sparse matrix compression

2002 ◽  
Author(s):  
S.G. Nastea ◽  
T. El-Ghazawi ◽  
O. Frieder
Keyword(s):  
Sparse Matrix ◽  
Input Output ◽  
Matrix Compression ◽  
Parallel Input

Expansion of Ratio Range of Shaft Drive CVT by Using Zero-Spin Disk

2007 ◽  
Author(s):  
Yukihito Narita ◽  
Masashi Yamanaka ◽  
Katsumi Inoue
Keyword(s):  
Slip Rate ◽  
Speed Ratio ◽  
The Novel ◽  
Evaluation Function ◽  
Input Output ◽  
Traction Drive ◽  
Parallel Input ◽  
Ratio Range

The novel mechanism CVT (Shaft Drive CVT, S-CVT) was developed by the authors. It transmits power by a traction drive same as the half toroidal CVT. S-CVT has parallel input/output shafts with conical or concave disks and the idler shaft having conical rollers at both ends, which is placed perpendicularly to the input/output shafts. All disks and rollers can move along each axis directions, and these movements produce the ratio changing by the changes of the rotational radii. The efficiency is the key evaluation function of CVT. To improve the efficiency, the backup roller mechanism was devised. Its effectiveness was confirmed by the experiment, and the efficiency of 95% was obtained by modified prototype S-CVT. This paper deals with the expansion of ratio range of S-CVT. In case of using the present disks, S-CVT has a difficulty to expand the narrow ratio range of 4 (0.5 to 2) because of the large slip brought by the spin. To expand the ratio range, the zero-spin disk/roller was devised. The shape of zero-spin disk/roller satisfies the condition that the spin does not occur at any speed ratio. According to the calculation, the slip rate becomes less than 1% at any speed ratio. To confirm the effectiveness, the prototype S-CVT with zero-spin disk was manufactured. It has the ratio range of 0.43 to 2.35. To obtain the slip rate the experiments were carried out at the speed ratio of 0.43, 1 and 2.35. At each speed ratio, the slip rate of less than 1% was obtained, and the effect of the zerospin disk was confirmed.

Machine Learning to Design an Auto-tuning System for the Best Compressed Format Detection for Parallel Sparse Computations

2021 ◽  
Author(s):  
Olfa Hamdi-Larbi ◽  
Ichrak Mehrez ◽  
Thomas Dufaud
Keyword(s):  
Machine Learning ◽  
Numerical Method ◽  
High Performance ◽  
Programming Model ◽  
Learning Algorithm ◽  
Sparse Matrix ◽  
Sparse Matrices ◽  
Matrix Compression ◽  
Target Architecture ◽  
Parallel Programming Model

Many applications in scientific computing process very large sparse matrices on parallel architectures. The presented work in this paper is a part of a project where our general aim is to develop an auto-tuner system for the selection of the best matrix compression format in the context of high-performance computing. The target smart system can automatically select the best compression format for a given sparse matrix, a numerical method processing this matrix, a parallel programming model and a target architecture. Hence, this paper describes the design and implementation of the proposed concept. We consider a case study consisting of a numerical method reduced to the sparse matrix vector product (SpMV), some compression formats, the data parallel as a programming model and, a distributed multi-core platform as a target architecture. This study allows extracting a set of important novel metrics and parameters which are relative to the considered programming model. Our metrics are used as input to a machine-learning algorithm to predict the best matrix compression format. An experimental study targeting a distributed multi-core platform and processing random and real-world matrices shows that our system can improve in average up to 7% the accuracy of the machine learning.

Enhance parallel input/output with cross-bundle aggregation

2015 ◽  
Vol 30 (2) ◽  
pp. 241-256
Author(s):  
Teng Wang ◽  
Kevin Vasko ◽  
Zhuo Liu ◽  
Hui Chen ◽  
Weikuan Yu
Keyword(s):  
Input Output ◽  
Parallel Input

Streaming Sparse Matrix Compression/Decompression

2005 ◽  
pp. 116-129 ◽  
Author(s):  
David Moloney ◽  
Dermot Geraghty ◽  
Colm McSweeney ◽  
Ciaran McElroy
Keyword(s):  
Sparse Matrix ◽  
Matrix Compression

scholarly journals A graph-theoretic approach to sparse matrix inversion for implicit differential algebraic equations

2013 ◽  
Vol 4 (1) ◽  
pp. 243-250
Author(s):  
H. Yoshimura
Keyword(s):  
Bipartite Graph ◽  
Large Scale ◽  
Jacobian Matrix ◽  
Sparse Matrix ◽  
Mechanical Systems ◽  
Differential Algebraic Equations ◽  
Matrix Inversion ◽  
Theoretic Approach ◽  
Algebraic Equations ◽  
Input Output

Abstract. In this paper, we propose an efficient numerical scheme to compute sparse matrix inversions for Implicit Differential Algebraic Equations of large-scale nonlinear mechanical systems. We first formulate mechanical systems with constraints by Dirac structures and associated Lagrangian systems. Second, we show how to allocate input-output relations to the variables in kinematical and dynamical relations appearing in DAEs by introducing an oriented bipartite graph. Then, we also show that the matrix inversion of Jacobian matrix associated to the kinematical and dynamical relations can be carried out by using the input-output relations and we explain solvability of the sparse Jacobian matrix inversion by using the bipartite graph. Finally, we propose an efficient symbolic generation algorithm to compute the sparse matrix inversion of the Jacobian matrix, and we demonstrate the validity in numerical efficiency by an example of the stanford manipulator.

Sign in / Sign up

Export Citation Format

Share Document