Robotique, Image Et Vision Scalable Parallel Geometric Algorithms for Coarse Grained Multicomputers Scalable Parallel Geometric Algorithms for Coarse Grained Multicomputers Algorithmes G Eom Etriques Parall Eles Multi-echelles Pour Machines Multi-processeurs a Gros Grain
نویسندگان
چکیده
Whereas most of the literature assumes that the number of processors p is a function of the problem size n, in scalable algorithms p becomes a parameter of the time complexity. This is a more realistic modelisation of real parallel machines and yields optimal algorithms, for the case that n p, where \ " is a function depending on the architecture of the interconnection network. In this paper we present scalable algorithms for a number of geometric problems, namely lower envelope of line segments, 2D-nearest neighbour, 3D-maxima, 2D-weighted dominance counting, area of the union of rectangles, 2D-convex hull. The main idea of these algorithms is to decompose the problem in p subproblems of sizeO(np + f(p)), with f(p) np , which can be solved independently using optimal sequential algorithms. For each problem we present a spatial decomposition scheme based on some geometric observations. The decomposition schemes have in common that they can be computed by globally sorting the entire data set at most twice. The data redundancy of f(p) duplicates of data elements per processor does not increase the asymptotic time complexity, and ranges, for the algorithms presented in this paper, from p to p2. The algorithms do not depend on a speci c architecture, they are easy to implement and in practice e cient as experiments show. R esum e Tandis que la plupart de la litt erature suppose que le nombre de processeurs p est une fonction de la taille du probl eme n, dans les algorithmes multiechelles p devient un param etre de la complexit e en temps. Ceci est une mod elisation plus r ealiste de machines parall eles r eelles et donne des algorithmes optimaux, dans le cas n p, o u \ " est une fonction d ependant de l'architecture du r eseau d'interconnection. Dans cet article nous pr esentons des algorithmes multiechelles pour plusieurs probl emes g eom etriques, l'enveloppe inf erieure de segments de droites, plus proche voisin dans le plan, maxima 3D, comptage de la dominance a poids, aire de l'union de rectangles, enveloppe convexe dans le plan. L`id ee principale de ces algorithmes consiste a d ecomposer le probl eme en p sous-probl emes de taille O(np + f(p)), avec f(p) np , qui peuvent être r esolus ind ependamment en utilisant des algorithmes s equentiels optimaux. Pour chaque probl eme nous pr esentons un sch ema de d ecomposition spatiale bas e sur des observations g eom etriques. Les sch emas de d ecomposition ont en commun le fait qu'ils peuvent être calcul es en triant l'ensemble entier de donn ees au plus deux fois. La redondance de f(p) copies de donn ees par processeur n'augmente pas la complexit e asymptotique en temps, et varie, pour les algorithmes pr esent es dans cet article, entre p et p2. Les algorithmes ne d ependent pas d'une architecture sp eci que, ils sont simples a impl ementer et e caces en pratique, comme des r esultats exp erimentaux le montrent.
منابع مشابه
The Handling of Graphs on PC Clusters: A Coarse Grained Approach
We study the relationship between the design and analysis of graph algorithms in the coarsed grained parallel models and the behavior of the resulting code on clusters. We conclude that the coarse grained multicomputer model (CGM) is well suited to design competitive algorithms, and that it is thereby now possible to aim to develop portable, predictable and efficient parallel code for graph pro...
متن کاملD-dimensional Range Search on Multicomputers Ecole Normale Supérieure De Lyon D-dimensional Range Search on Multicomputers D-dimensional Range Search on Multicomputers
Given a set L of n points in the d-dimensional Cartesian space E d , and a query specifying a domain q in E d , the Range Search problem consists in identifying the subset R(q) of the points of L contained in q. The Range Tree data-structure represents a particularly good balance between storage space and search time. The structure requires O(n log d?1 n) space and construction time, but suppor...
متن کاملGraph coloring on coarse grained multicomputers
We present an efficient and scalable Coarse Grained Multicomputer (CGM) coloring algorithm that colors a graph G with at most ∆+1 colors where ∆ is the maximum degree in G. This algorithm is given in two variants: a randomized and a deterministic. We show that on a p-processor CGM model the proposed algorithms require a parallel time of O( |G| p ) and a total work and overall communication cost...
متن کاملScalable Coarse Grained Parallel Interval Graph Algorithms
We present scalable coarse grained parallel algorithms for solving interval graph problems on a BSP-like model{Coarse Grained Multicomputers (CGM). The problems we consider include: nding maximum independent set, maximum weighted clique, minimum coloring and cut vertices and bridges. With scalability at n p p ; 8 > 0 (here n denotes the total input size and p the number of processors), our algo...
متن کاملThe Saukas-Song Selection Algorithm and Coarse Grained Parallel Sorting
We analyze the running time of the Saukas-Song algorithm for selection on a coarse grained multicomputer without expressing the running time in terms of communication rounds. We derive sorting algorithms that are asymptotically optimal for restricted ranges of processors on coarse grained multicomputers.
متن کامل