Efficient broadcasts and simple algorithms for parallel linear algebra computing in clusters

Detalles Bibliográficos
Autor Principal: Tinetti, Fernando Gustavo
Otros autores o Colaboradores: Luque, Emilio
Formato: Capítulo de libro
Lengua:inglés
Series:^p Datos electrónicos (1 archivo : 246 KB)
Temas:
Acceso en línea:ieeexplore.ieee.org/stamp/stamp.jsp?arnumber1213364&isnumber27277
Consultar en el Cátalogo
Resumen:This paper presents a natural and efficient implementation for the classical broadcast message passing routine which optimizes performance of Ethernet based clusters. A simple algorithm for parallel matrix multiplication is specifically designed to take advantage of both, parallel computing facilities (CPUs) provided by clusters, and optimized performance of broadcast messages on Ethernet based clusters. Also, this simple parallel algorithm proposed for matrix multiplication takes into account the possibly heterogeneous computing hardware and maintains a balanced workload of computers according to their relative computing power. Performance tests are presented on a heterogeneous cluster as well as on a homogeneous cluster, where it is compared with the parallel matrix multiplication provided by the ScaLAPACK library. Another simple parallel algorithm is proposed for LU matrix factorization (a general method to solve dense systems of equations) following the same guidelines used for the parallel matrix multiplication algorithm. Some performance tests are presented over a homogeneous cluster.
Notas:Formato de archivo: PDF. -- Este documento es producción intelectual de la Facultad de Informática-UNLP (Colección BIPA / Biblioteca.) -- Disponible también en línea (Cons. 13/03/2009)

MARC

LEADER 00000naa a2200000 a 4500
003 AR-LpUFIB
005 20250423183003.0
008 230201s2003 xx o 000 0 eng d
024 8 |a DIF-M2650  |b 2742  |z DIF002556 
040 |a AR-LpUFIB  |b spa  |c AR-LpUFIB 
100 1 |a Tinetti, Fernando Gustavo  |9 44771 
245 1 0 |a Efficient broadcasts and simple algorithms for parallel linear algebra computing in clusters 
490 0 |a ^p Datos electrónicos (1 archivo : 246 KB) 
500 |a Formato de archivo: PDF. -- Este documento es producción intelectual de la Facultad de Informática-UNLP (Colección BIPA / Biblioteca.) -- Disponible también en línea (Cons. 13/03/2009) 
520 |a This paper presents a natural and efficient implementation for the classical broadcast message passing routine which optimizes performance of Ethernet based clusters. A simple algorithm for parallel matrix multiplication is specifically designed to take advantage of both, parallel computing facilities (CPUs) provided by clusters, and optimized performance of broadcast messages on Ethernet based clusters. Also, this simple parallel algorithm proposed for matrix multiplication takes into account the possibly heterogeneous computing hardware and maintains a balanced workload of computers according to their relative computing power. Performance tests are presented on a heterogeneous cluster as well as on a homogeneous cluster, where it is compared with the parallel matrix multiplication provided by the ScaLAPACK library. Another simple parallel algorithm is proposed for LU matrix factorization (a general method to solve dense systems of equations) following the same guidelines used for the parallel matrix multiplication algorithm. Some performance tests are presented over a homogeneous cluster. 
534 |a Parallel and Distributed Processing Symposium, 2003. 
650 4 |a PROCESAMIENTO PARALELO  |9 42834 
650 4 |a ALGORITMOS PARALELOS  |9 42911 
650 4 |a CLUSTERS  |9 44129 
650 4 |a ÁLGEBRA LINEAL  |9 44372 
650 4 |a REDES LOCALES  |9 46432 
650 4 |a INTERCONEXIÓN DE REDES  |9 43257 
650 4 |a COMUNICACIÓN DE DATOS  |9 42893 
700 1 |a Luque, Emilio  |9 46169 
856 4 0 |u ieeexplore.ieee.org/stamp/stamp.jsp?arnumber1213364&isnumber27277 
942 |c CP 
952 |0 0  |1 0  |4 0  |6 A0148  |7 3  |8 BD  |9 76998  |a DIF  |b DIF  |d 2025-03-11  |l 0  |o A0148  |r 2025-03-11 17:02:47  |u http://catalogo.info.unlp.edu.ar/meran/getDocument.pl?id=154  |w 2025-03-11  |y CP 
999 |c 52435  |d 52435