Back to Search
Start Over
Use of A Network Enabled Server System for a Sparse Linear Algebra Application
- Source :
- Workshop on Future Generation Grids, Workshop on Future Generation Grids, Nov 2006, Dagstuhl, Germany. pp.171-193, ⟨10.1007/978-0-387-29445-2_10⟩, Future Generation Grids ISBN: 9780387279350, Future Generation Grids
- Publication Year :
- 2006
- Publisher :
- HAL CCSD, 2006.
-
Abstract
- International audience; Solving systems of linear equations is one of the key operations in linear algebra. Many different algorithms are available in that purpose. These algorithms require a very accurate tuning to minimise runtime and memory consumption. The TLSE project provides, on one hand, a scenario-driven expert site to help users choose the right algorithm according to their problem and tune accurately this algorithm, and, on the other hand, a test-bed for experts in order to compare algorithms and define scenarios for the expert site. Both features require to run the available solvers a large number of times with many different values for the control parameters (and maybe with many different architectures). Currendy, only the grid can provide enough computing power for this kind of application. The DIET middleware is the GRID backbone for TLSE. It manages the solver services and their scheduling in a scalable way.
- Subjects :
- Computer science
Distributed computing
Grid application
[INFO.INFO-OH]Computer Science [cs]/Other [cs.OH]
010103 numerical & computational mathematics
02 engineering and technology
Solver
computer.software_genre
Grid
System of linear equations
01 natural sciences
Scheduling (computing)
Grid computing
Linear algebra
Scalability
0202 electrical engineering, electronic engineering, information engineering
020201 artificial intelligence & image processing
0101 mathematics
computer
Subjects
Details
- Language :
- English
- ISBN :
- 978-0-387-27935-0
- ISBNs :
- 9780387279350
- Database :
- OpenAIRE
- Journal :
- Workshop on Future Generation Grids, Workshop on Future Generation Grids, Nov 2006, Dagstuhl, Germany. pp.171-193, ⟨10.1007/978-0-387-29445-2_10⟩, Future Generation Grids ISBN: 9780387279350, Future Generation Grids
- Accession number :
- edsair.doi.dedup.....dcdee1b2f1b677324f63b90df97f67c9
- Full Text :
- https://doi.org/10.1007/978-0-387-29445-2_10⟩