A MATLAB-Based Code Generator for Sparse Matrix Computations

Kawabata, Hideyuki; Suzuki, Mutsumi; Kitamura, Toshiaki

doi:10.1007/978-3-540-30477-7_19

Cited by 6 publications

(8 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…User programs for these systems are written in a manner similar to the traditional function calls, while user programs in SILC make use of mathematical expressions to request matrix computations. Telescoping Languages [5] and CMC [4] are source-to-source code translation systems, in which user programs are written in the MATLAB language [11]. The focus of Telescoping Languages is on the optimization of both user programs and matrix computation libraries, whereas CMC focuses on the generation of optimized Fortran subroutines from user programs as well as on rich support for various sparse matrix storage formats.…”

Section: Replacement Of Matrix Computation Libraries Is Requiredmentioning

confidence: 99%

LAPACK in SILC: use of a flexible application framework for matrix computation libraries

Kajiyama

Nukada

Suda

et al. 2005

Eighth International Conference on High-Performance Computing in Asia-Pacific Region (HPCASIA'05)

View full text Add to dashboard Cite

This paper presents a novel application framework named Simple Interface for Library Collections (SILC) that allows users to make use of matrix computation libraries in a flexible and language-independent manner. Using SILC, various computing environments as well as alternative solvers and matrix storage formats from different libraries can be easily utilized. The present paper describes the design and implementation of SILC for shared-memory parallel computing environments, and discusses the use of LAPACK in the framework of SILC together with some experimental results on the performance of the implemented system.

show abstract

Section: Replacement Of Matrix Computation Libraries Is Requiredmentioning

confidence: 99%

LAPACK in SILC: use of a flexible application framework for matrix computation libraries

Kajiyama

Nukada

Suda

et al. 2005

Eighth International Conference on High-Performance Computing in Asia-Pacific Region (HPCASIA'05)

View full text Add to dashboard Cite

show abstract

“…Our prototype of MATLAB-to-PSBLAS translator is based on CMC which translates MATLAB-based scripts into Fortran [14], [15]. The flow of the translation by CMC is illustrated in Fig.…”

Section: Analysis Of Matlab Scripts By Cmcmentioning

confidence: 99%

“…For operations + and -, resulting shape is F when the scalar value is not zero. code motion, and common subexpression elimination, and matrix-computation-specific ones like strength reduction of matrix operations are performed [14], [15]. In some cases, dynamic memory management mechanisms have to be incorporated.…”

Section: Analysis Of Matlab Scripts By Cmcmentioning

confidence: 99%

“…Figure 1 clearly depicts the similarity of the syntax of both descriptions. MATLAB's execution environment is interpreter-based and its performance is known to be problematic in some cases, but language translation techniques exist to make MATLAB programs run fast [11]- [14] and some of them support generating parallel codes. However, as for parallel and sparse computations, there is no programming environment available to the authors' knowledge.…”

Section: Introductionmentioning

confidence: 99%

“…The system generates subroutines for an SPMD-style parallel program which runs on machines where PSBLAS library is available, i.e., most parallel computers with MPI communication library. The functionality of generating PSBLAS code is being built into our matrix language compiler named CMC which translates MATLAB-based scripts into Fortran or C [14], [15]. Experimental results show that our prototype system can emit fairly efficient PSBLAS programs for simple linear sparse solvers.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A MATLAB-Based Code Generator for Parallel Sparse Matrix Computations Utilizing PSBLAS

Sasaoka¹,

Kawabata²,

Kitamura³

2007

IEICE Transactions on Information and Systems

View full text Add to dashboard Cite

Parallel programs for distributed memory machines are not easy to create and maintain, especially when they involve sparse matrix computations. In this paper, we propose a program translation system for generating parallel sparse matrix computation codes utilizing PS-BLAS. The purpose of the development of the system is to offer the user a convenient way to construct parallel sparse code based on PSBLAS. The system is build up on the idea of bridging the gap between the easy-toread program representations and highly-tuned parallel executables based on existing parallel sparse matrix computation libraries. The system accepts a MATLAB program with annotations and generates subroutines for an SPMD-style parallel program which runs on distributed-memory machines. Experimental results on parallel machines show that the prototype of our system can generate fairly efficient PSBLAS codes for simple applications such as CG and Bi-CGSTAB programs. key words: matrix computations, MATLAB, program translation, parallel libraries * Note that Fortran 90's matrix operations are not consistent with mathematical notations-you can not write A*B when both A and B are not scalar variables.

show abstract

SILC: A Flexible and Environment-Independent Interface for Matrix Computation Libraries

Kajiyama

Nukada

Hasegawa

et al. 2006

Parallel Processing and Applied Mathematics

View full text Add to dashboard Cite

A MATLAB-Based Code Generator for Sparse Matrix Computations

Cited by 6 publications

References 9 publications

LAPACK in SILC: use of a flexible application framework for matrix computation libraries

LAPACK in SILC: use of a flexible application framework for matrix computation libraries

A MATLAB-Based Code Generator for Parallel Sparse Matrix Computations Utilizing PSBLAS

SILC: A Flexible and Environment-Independent Interface for Matrix Computation Libraries

Contact Info

Product

Resources

About