-
M. Faverge.
Ordonnancement hybride statique-dynamique en algèbre linéaire creuse pour de grands clusters de machines NUMA et multi-coeurs.
PhD thesis,
LaBRI, Université Bordeaux I, Talence,
Talence, France,
December 2009.
Keyword(s): Sparse.
Abstract: |
New supercomputers incorporate many microprocessors which include themselves one or many computational cores. These new architectures induce strongly hierarchical topologies. These are called NUMA architectures. Sparse direct solvers are a basic building block of many numerical simulation algorithms. They need to be adapted to these new architectures with Non Uniform Memory Accesses. We propose to introduce a dynamic scheduling designed for NUMA architectures in the PaStiX, solver. The data structures of the solver, as well as the patterns of communication have been modified to meet the needs of these architectures and dynamic scheduling. We are also interested in the dynamic adaptation of the computation grain to use efficiently multi-core architectures and shared memory. Experiments on several numerical test cases will be presented to prove the efficiency of the approach on different architectures. |
@PhdThesis{t:LaBRI::MF09,
author = "Faverge, M.",
title = "{O}rdonnancement hybride statique-dynamique en algèbre linéaire creuse pour de grands clusters de machines {NUMA} et multi-coeurs",
school = "LaBRI, Université Bordeaux I, Talence",
year = "2009",
OPTkey = {},
address = {Talence, France},
OPTtype = {},
month = dec,
OPTnote = {},
OPTannote = {},
URL = {http://www.labri.fr/~ramet/restricted/these_faverge.pdf.gz},
KEYWORDS = "Sparse",
ABSTRACT = {New supercomputers incorporate many microprocessors which include themselves one or many computational cores. These new architectures induce strongly hierarchical topologies. These are called NUMA architectures. Sparse direct solvers are a basic building block of many numerical simulation algorithms. They need to be adapted to these new architectures with Non Uniform Memory Accesses. We propose to introduce a dynamic scheduling designed for NUMA architectures in the PaStiX, solver. The data structures of the solver, as well as the patterns of communication have been modified to meet the needs of these architectures and dynamic scheduling. We are also interested in the dynamic adaptation of the computation grain to use efficiently multi-core architectures and shared memory. Experiments on several numerical test cases will be presented to prove the efficiency of the approach on different architectures.}
}
-
J. Gaidamour.
Conception d'un solveur linéaire creux parallèle hybride direct-itératif.
PhD thesis,
LaBRI, Université Bordeaux I, Talence,
Talence, France,
December 2009.
Keyword(s): Sparse.
Abstract: |
This thesis presents a parallel resolution method for sparse linear systems which combines effectively techniques of direct and iterative solvers using a Schur complement approach. A domain decomposition is built ; the interiors of the subdomains are eliminated by a direct method in order to use an iterative method only on the interface unknowns. The system on the interface (Schur complement) is solved thanks to an iterative method preconditioned by a global incomplete factorization. A special ordering on the Schur complement allows to build a scalable preconditioner. Algorithms minimizing the memory peak that appears during the construction of the preconditioner are presented. The memory is balanced thanks to a multiple domains per processors parallelization scheme. The methods are implemented in the Hips solver and parallel experimental results are presented on large industrial test cases. |
@PhdThesis{t:LaBRI::JG09,
author = "Gaidamour, J.",
title = "Conception d'un solveur lin\'eaire creux parall\`ele hybride direct-it\'eratif",
school = "LaBRI, Universit\'e Bordeaux I, Talence",
year = "2009",
OPTkey = {},
address = {Talence, France},
OPTtype = {},
month = dec,
OPTnote = {},
OPTannote = {},
URL = {http://www.labri.fr/~ramet/restricted/these_gaidamour.pdf.gz},
KEYWORDS = "Sparse",
ABSTRACT = {This thesis presents a parallel resolution method for sparse linear systems which combines effectively techniques of direct and iterative solvers using a Schur complement approach. A domain decomposition is built ; the interiors of the subdomains are eliminated by a direct method in order to use an iterative method only on the interface unknowns. The system on the interface (Schur complement) is solved thanks to an iterative method preconditioned by a global incomplete factorization. A special ordering on the Schur complement allows to build a scalable preconditioner. Algorithms minimizing the memory peak that appears during the construction of the preconditioner are presented. The memory is balanced thanks to a multiple domains per processors parallelization scheme. The methods are implemented in the Hips solver and parallel experimental results are presented on large industrial test cases.}
}
-
M. Barrault,
B. Lathuilière,
P. Ramet,
and J. Roman.
A Non Overlapping Parallel Domain Decomposition Method Applied to The Simplified Transport Equations.
In International Conference on Mathematics, Computational Methods & Reactor Physics,
New-York, USA,
May 2009.
Keyword(s): DDM.
@InProceedings{C:LaBRI::neutron2009,
author = {Barrault, M. and Lathuili\`ere, B. and Ramet, P. and Roman, J.},
title = {A Non Overlapping Parallel Domain Decomposition Method Applied to The Simplified Transport Equations},
OPTcrossref = {},
OPTkey = {},
booktitle = {International Conference on Mathematics, Computational Methods & Reactor Physics},
OPTpages = {},
year = {2009},
OPTeditor = {},
OPTvolume = {},
OPTnumber = {},
OPTseries = {},
address = {New-York, USA},
month = may,
OPTorganization = {},
OPTpublisher = {},
URL = {http://www.labri.fr/~ramet/restricted/mc09.pdf.gz},
KEYWORDS = "DDM",
OPTannote = {}
}
-
M. Faverge.
A NUMA Aware Scheduler for a Parallel Sparse Direct Solver.
In Journées Informatique Massivement Multiprocesseur et Multicoeur,
Rocquencourt, France,
February 2009.
Keyword(s): Sparse.
@InProceedings{c:LaBRI::i3m,
author = {Faverge, M.},
title = {A NUMA Aware Scheduler for a Parallel Sparse Direct Solver},
booktitle = "Journées Informatique Massivement Multiprocesseur et Multicoeur",
OPTcrossref = {},
OPTkey = {},
OPTpages = {},
year = {2009},
OPTeditor = {},
OPTvolume = {},
OPTnumber = {},
OPTseries = {},
address = {Rocquencourt, France},
month = feb,
OPTorganization = {},
OPTpublisher = {},
OPTannote = {},
URL = {http://www.labri.fr/~ramet/restricted/i3m.pdf.gz},
KEYWORDS = "Sparse",
}
-
M. Faverge.
Dynamic Scheduling for Sparse Direct Solver on NUMA and Multicore Architectures.
In Sparse Days,
Toulouse, France,
June 2009.
Keyword(s): Sparse.
@InProceedings{C:LaBRI::sparsedays2009,
author = {Faverge, M.},
title = {Dynamic Scheduling for Sparse Direct Solver on NUMA and Multicore Architectures},
booktitle = "Sparse Days",
OPTcrossref = {},
OPTkey = {},
OPTpages = {},
year = {2009},
OPTeditor = {},
OPTvolume = {},
OPTnumber = {},
OPTseries = {},
address = {Toulouse, France},
month = jun,
OPTorganization = {},
OPTpublisher = {},
OPTannote = {},
KEYWORDS = "Sparse",
}
-
M. Faverge.
Vers un solveur de systèmes linéaires creux adapté aux machines NUMA.
In ACTES RenPar'2009,
Toulouse, France,
September 2009.
Keyword(s): Sparse.
@InProceedings{c:LaBRI::renpar09,
author = {Faverge, M.},
title = {Vers un solveur de systèmes linéaires creux adapté aux machines {NUMA}},
booktitle = "ACTES RenPar'2009",
OPTcrossref = {},
OPTkey = {},
OPTpages = {},
year = {2009},
OPTeditor = {},
OPTvolume = {},
OPTnumber = {},
OPTseries = {},
address = {Toulouse, France},
month = sep,
OPTorganization = {},
OPTpublisher = {},
OPTnote = {},
OPTannote = {},
URL = {http://www.labri.fr/~ramet/restricted/renpar09.ps.gz},
KEYWORDS = "Sparse",
}
-
G. Huysmans,
Pamela S.,
E. van der Plas,
and P. Ramet.
Non-Linear MHD simulations of Edge Localised Modes.
In 36th EPS Plasma Physics Conference,
Sofia, Bulgaria,
June 2009.
Keyword(s): Fusion.
@InProceedings{C:LaBRI::EPS2009,
author = {Huysmans, G. and Pamela S. and van~der~Plas, E. and Ramet, P.},
title = {Non-Linear {MHD} simulations of Edge Localised Modes},
OPTcrossref = {},
OPTkey = {},
booktitle = {36th {EPS} Plasma Physics Conference},
OPTpages = {},
year = {2009},
OPTeditor = {},
OPTvolume = {},
OPTnumber = {},
OPTseries = {},
address = {Sofia, Bulgaria},
month = jun,
OPTorganization = {},
OPTpublisher = {},
URL = {http://www.labri.fr/~ramet/restricted/eps09.pdf.gz},
KEYWORDS = "Fusion",
OPTannote = {}
}
-
P. Ramet.
Dynamic Scheduling for Sparse Direct Solver on NUMA and Multicore Architectures.
In ComplexHPC meeting,
Lisbon, Portugal,
October 2009.
Keyword(s): Sparse.
@InProceedings{C:LaBRI::COST2009,
author = {Ramet, P.},
title = {Dynamic Scheduling for Sparse Direct Solver on NUMA and Multicore Architectures},
booktitle = "ComplexHPC meeting",
OPTcrossref = {},
OPTkey = {},
OPTpages = {},
year = {2009},
OPTeditor = {},
OPTvolume = {},
OPTnumber = {},
OPTseries = {},
address = {Lisbon, Portugal},
month = oct,
OPTorganization = {},
OPTpublisher = {},
OPTannote = {},
KEYWORDS = "Sparse",
}