Performance model for parallel mathematical libraries based on historical knowledgebase

I. Salawdeh; E. César; A. Morajko; T. Margalef; E. Luque

doi:10.1007/978-3-540-85451-7_13

Performance model for parallel mathematical libraries based on historical knowledgebase

I. Salawdeh^*, E. César, A. Morajko, T. Margalef, E. Luque

^*Corresponding author for this work

Department of Computer Architecture and Operating Systems

Research output: Contribution to journal › Article › Research › peer-review

1 Citation (Scopus)

Abstract

Scientific and mathematical parallel libraries offer a high level of abstraction to programmers. However, it is still difficult to select the proper parameters and algorithms to maximize the application performance. This work proposes a performance model for dynamically adjusting applications written with the PETSc library. This model is based on historical performance information and data mining techniques. Finally, we demonstrate the validity of the proposed model through real experimentations.

Original language	American English
Pages (from-to)	110-119
Number of pages	10
Journal	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
DOIs	https://doi.org/10.1007/978-3-540-85451-7_13
Publication status	Published - 2008

Keywords

Dynamic mathematical model
Mathematical Performance
Performance Model
PETSc Performance

Access to Document

10.1007/978-3-540-85451-7_13

Link to publication in Scopus

Fingerprint Dive into the research topics of 'Performance model for parallel mathematical libraries based on historical knowledgebase'. Together they form a unique fingerprint.

Cite this

@article{8891b24c0ba34349a095bdf452487724,

title = "Performance model for parallel mathematical libraries based on historical knowledgebase",

abstract = "Scientific and mathematical parallel libraries offer a high level of abstraction to programmers. However, it is still difficult to select the proper parameters and algorithms to maximize the application performance. This work proposes a performance model for dynamically adjusting applications written with the PETSc library. This model is based on historical performance information and data mining techniques. Finally, we demonstrate the validity of the proposed model through real experimentations.",

keywords = "Dynamic mathematical model, Mathematical Performance, Performance Model, PETSc Performance",

author = "I. Salawdeh and E. C{\'e}sar and A. Morajko and T. Margalef and E. Luque",

year = "2008",

doi = "10.1007/978-3-540-85451-7_13",

language = "Ingl{\'e}s estadounidense",

pages = "110--119",

}

Performance model for parallel mathematical libraries based on historical knowledgebase. / Salawdeh, I.; César, E.; Morajko, A.; Margalef, T.; Luque, E.

In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2008, p. 110-119.

Research output: Contribution to journal › Article › Research › peer-review

TY - JOUR

T1 - Performance model for parallel mathematical libraries based on historical knowledgebase

AU - Salawdeh, I.

AU - César, E.

AU - Morajko, A.

AU - Margalef, T.

AU - Luque, E.

PY - 2008

Y1 - 2008

N2 - Scientific and mathematical parallel libraries offer a high level of abstraction to programmers. However, it is still difficult to select the proper parameters and algorithms to maximize the application performance. This work proposes a performance model for dynamically adjusting applications written with the PETSc library. This model is based on historical performance information and data mining techniques. Finally, we demonstrate the validity of the proposed model through real experimentations.

AB - Scientific and mathematical parallel libraries offer a high level of abstraction to programmers. However, it is still difficult to select the proper parameters and algorithms to maximize the application performance. This work proposes a performance model for dynamically adjusting applications written with the PETSc library. This model is based on historical performance information and data mining techniques. Finally, we demonstrate the validity of the proposed model through real experimentations.

KW - Dynamic mathematical model

KW - Mathematical Performance

KW - Performance Model

KW - PETSc Performance

UR - http://www.scopus.com/inward/record.url?scp=51849090611&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-85451-7_13

DO - 10.1007/978-3-540-85451-7_13

M3 - Artículo

AN - SCOPUS:51849090611

SP - 110

EP - 119

ER -