J. Aliaga, F. Almeida, J. M. Badía, S. Barrachina, Vicente Blanco Pérez, María Isabel Castillo, U. Dorta, R. Mayo, E. S. Quintana‐Ortí, G. Quintana-Ortí, C. Rodríguez, F. D. Sande
In this paper we present our joint efforts towards the development of a parallel version of the GNU Scientific Library for heterogeneous systems. Two well-known operations arising in discrete mathematics and sparse linear algebra allow us to describe the design and the implementation of the library, and to report experimental results on heterogeneous clusters of personal computers.
{"title":"Parallelization of the GNU Scientific Library on heterogeneous systems","authors":"J. Aliaga, F. Almeida, J. M. Badía, S. Barrachina, Vicente Blanco Pérez, María Isabel Castillo, U. Dorta, R. Mayo, E. S. Quintana‐Ortí, G. Quintana-Ortí, C. Rodríguez, F. D. Sande","doi":"10.1109/ISPDC.2004.39","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.39","url":null,"abstract":"In this paper we present our joint efforts towards the development of a parallel version of the GNU Scientific Library for heterogeneous systems. Two well-known operations arising in discrete mathematics and sparse linear algebra allow us to describe the design and the implementation of the library, and to report experimental results on heterogeneous clusters of personal computers.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82848770","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
K. Ferreira, S. Yamagiwa, L. Sousa, K. Aoki, K. Wada, L. M. Campos
Cluster computing has become a valid alternative for high performance computing. To fully exploit the computing power of these environments, one must utilize high performance network and protocol technologies, since parallel applications running on clusters require low latency and high throughput. To address this issue, Maestro2 high performance network technology has been developed. Parallel applications running on clusters usually use two different known alternatives to share information; message passing or distributed shared memory. Maestro2 already supports a high performance message passing communication system, MMP. This paper describes a first version of a distributed shared memory system based on Maestro2, and a first set of experimental results is presented.
{"title":"Distributed shared memory system based on the Maestro2 high performance cluster network","authors":"K. Ferreira, S. Yamagiwa, L. Sousa, K. Aoki, K. Wada, L. M. Campos","doi":"10.1109/ISPDC.2004.19","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.19","url":null,"abstract":"Cluster computing has become a valid alternative for high performance computing. To fully exploit the computing power of these environments, one must utilize high performance network and protocol technologies, since parallel applications running on clusters require low latency and high throughput. To address this issue, Maestro2 high performance network technology has been developed. Parallel applications running on clusters usually use two different known alternatives to share information; message passing or distributed shared memory. Maestro2 already supports a high performance message passing communication system, MMP. This paper describes a first version of a distributed shared memory system based on Maestro2, and a first set of experimental results is presented.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83800023","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The paper is devoted to scalability analysis of a typical linear algebra algorithm on heterogeneous clusters. We proof that traditional scalability metrics proposed for analysis of linear algebra algorithms is applicable on heterogeneous platform and investigate influence of three heterogeneous strategies of computation distribution to scalable universal matrix multiplication algorithm (SUMMA) scalability.
{"title":"Scalability analysis of matrix-matrix multiplication on heterogeneous clusters","authors":"A. Kalinov","doi":"10.1109/ISPDC.2004.45","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.45","url":null,"abstract":"The paper is devoted to scalability analysis of a typical linear algebra algorithm on heterogeneous clusters. We proof that traditional scalability metrics proposed for analysis of linear algebra algorithms is applicable on heterogeneous platform and investigate influence of three heterogeneous strategies of computation distribution to scalable universal matrix multiplication algorithm (SUMMA) scalability.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87375449","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Classical task scheduling employs a very simplified model of the target parallel system. Experiments demonstrated that this leads to inaccurate and inefficient schedules. Contention aware scheduling heuristics take the contention for communication resources into account, which improves the schedules significantly. Yet, one aspect remains to be investigated: the involvement of the processors in communication. This paper proposes a new scheduling model, called involvement-contention model, that integrates the consideration for the processor involvement into task scheduling. A list scheduling based heuristic is proposed for the new model, which produces significantly more accurate and efficient schedules in experiments on real parallel systems.
{"title":"Task scheduling: considering the processor involvement in communication","authors":"O. Sinnen, L. Sousa","doi":"10.1109/ISPDC.2004.48","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.48","url":null,"abstract":"Classical task scheduling employs a very simplified model of the target parallel system. Experiments demonstrated that this leads to inaccurate and inefficient schedules. Contention aware scheduling heuristics take the contention for communication resources into account, which improves the schedules significantly. Yet, one aspect remains to be investigated: the involvement of the processors in communication. This paper proposes a new scheduling model, called involvement-contention model, that integrates the consideration for the processor involvement into task scheduling. A list scheduling based heuristic is proposed for the new model, which produces significantly more accurate and efficient schedules in experiments on real parallel systems.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73222098","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Secure job submission across multiple virtual organisations becomes more important as Grids proliferate. WebCom-G will bring together the condensed graph model of computing and existing Grid software from the European DataGrid (EDG) project to allow users to execute complex tasks involving multiple Grids. In this paper we discuss the security aspects of a system to allow users of existing Grids to securely execute condensed graphs containing Grid jobs. We outline the process that bridges the EDG security architecture and Secure WebCom to make this possible.
{"title":"Bridging Secure WebCom and European DataGrid security for multiple VOs over multiple grids","authors":"D. O'Callaghan, B. Coghlan","doi":"10.1109/ISPDC.2004.14","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.14","url":null,"abstract":"Secure job submission across multiple virtual organisations becomes more important as Grids proliferate. WebCom-G will bring together the condensed graph model of computing and existing Grid software from the European DataGrid (EDG) project to allow users to execute complex tasks involving multiple Grids. In this paper we discuss the security aspects of a system to allow users of existing Grids to securely execute condensed graphs containing Grid jobs. We outline the process that bridges the EDG security architecture and Secure WebCom to make this possible.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73388371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Olivier Beaumont, Arnaud Legrand, L. Marchal, Y. Robert
In this paper, we consider steady-state scheduling techniques for mapping a collection of task graphs onto heterogeneous systems, such as clusters and grids. We advocate the use of steady-state scheduling to solve this difficult problem. Due to space limitations, we concentrate on complexity results. We show that the problem of optimizing the steady-state throughput is NP-complete in the general case. We formulate a compact version of the problem that belongs to the NP complexity class but which does not restrict the optimality of the solution. We provide many positive results in the extended version (Beaumont et al., 2004). Indeed, we show how to determine in polynomial time the best steady-state scheduling strategy for a large class of application graphs and for an arbitrary platform graphs, using a linear programming approach.
在本文中，我们考虑了将一组任务图映射到异构系统(如集群和网格)上的稳态调度技术。我们提倡使用稳态调度来解决这一难题。由于篇幅限制，我们主要关注复杂性结果。我们证明了在一般情况下，优化稳态吞吐量的问题是np完全的。我们提出了这个问题的一个精简版本，它属于NP复杂度类，但不限制解的最优性。我们在扩展版中提供了许多积极的结果(Beaumont et al.， 2004)。实际上，我们展示了如何使用线性规划方法在多项式时间内确定大型应用程序图和任意平台图的最佳稳态调度策略。
{"title":"Assessing the impact and limits of steady-state scheduling for mixed task and data parallelism on heterogeneous platforms","authors":"Olivier Beaumont, Arnaud Legrand, L. Marchal, Y. Robert","doi":"10.1109/ISPDC.2004.12","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.12","url":null,"abstract":"In this paper, we consider steady-state scheduling techniques for mapping a collection of task graphs onto heterogeneous systems, such as clusters and grids. We advocate the use of steady-state scheduling to solve this difficult problem. Due to space limitations, we concentrate on complexity results. We show that the problem of optimizing the steady-state throughput is NP-complete in the general case. We formulate a compact version of the problem that belongs to the NP complexity class but which does not restrict the optimality of the solution. We provide many positive results in the extended version (Beaumont et al., 2004). Indeed, we show how to determine in polynomial time the best steady-state scheduling strategy for a large class of application graphs and for an arbitrary platform graphs, using a linear programming approach.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74923140","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nowadays, mobile cellular networks are based on the well-known hexagonal mesh in which nodes are considered as base stations. However, this model does not really take into account the irregularity and the randomness of central offices locations in a large telephone network. As we cannot transform this telephone network into a regular structure, because of spatial variations of population density, consumer demand, other geographical and technological factors, Voronoi cellular networks appears to be the more suitable model. In this paper, we describe the Voronoi cellular model and its topological properties.
{"title":"Voronoi mobile cellular networks: topological properties","authors":"Anne-Elisabeth Baert, D. Semé","doi":"10.1109/ISPDC.2004.58","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.58","url":null,"abstract":"Nowadays, mobile cellular networks are based on the well-known hexagonal mesh in which nodes are considered as base stations. However, this model does not really take into account the irregularity and the randomness of central offices locations in a large telephone network. As we cannot transform this telephone network into a regular structure, because of spatial variations of population density, consumer demand, other geographical and technological factors, Voronoi cellular networks appears to be the more suitable model. In this paper, we describe the Voronoi cellular model and its topological properties.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82346261","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Grids enable the sharing of resources and problem solving in multi-organizational environments. The problem of resource management in such systems is very complex because the participants, resource owners and users, have their own requirements and objectives that need to be considered when making allocation decisions. To address this issue novel protocols that take into account the self-interest and incentives of these participants need to be developed. These kinds of protocols in which the participants maximize their own utilities only if they report their true parameters and follow the rules are called strategyproof protocols. In this paper we propose AGORA, an architecture for strategyproof computing in grids and present several strategyproof mechanisms for resource allocation that can be deployed on this architecture.
{"title":"AGORA: an architecture for strategyproof computing in grids","authors":"Daniel Grosu","doi":"10.1109/ISPDC.2004.9","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.9","url":null,"abstract":"Grids enable the sharing of resources and problem solving in multi-organizational environments. The problem of resource management in such systems is very complex because the participants, resource owners and users, have their own requirements and objectives that need to be considered when making allocation decisions. To address this issue novel protocols that take into account the self-interest and incentives of these participants need to be developed. These kinds of protocols in which the participants maximize their own utilities only if they report their true parameters and follow the rules are called strategyproof protocols. In this paper we propose AGORA, an architecture for strategyproof computing in grids and present several strategyproof mechanisms for resource allocation that can be deployed on this architecture.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90432759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
New parallel Normalized Explicit Preconditioned Conjugate Gradient-type methods are introduced for solving finite element systems on distributed memory MIMD systems. The performance and applicability of the proposed methods implemented in message passing interface, is discussed by solving a characteristic two dimensional boundary value problem and numerical results are given.
{"title":"Parallel approximate finite element inverse preconditioning on distributed systems","authors":"G. Gravvanis, K. M. Giannoutakis","doi":"10.1109/ISPDC.2004.36","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.36","url":null,"abstract":"New parallel Normalized Explicit Preconditioned Conjugate Gradient-type methods are introduced for solving finite element systems on distributed memory MIMD systems. The performance and applicability of the proposed methods implemented in message passing interface, is discussed by solving a characteristic two dimensional boundary value problem and numerical results are given.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85063554","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this work, we propose to extend the standard concept of planar cellular networks into space. Indeed, in cellular networks the trend is to have smaller cells for indoor environment. Then, we give a description of a 3D picocellular network based on layers of hexagonal mesh to ensure that challenge.
{"title":"A new model for 3D cellular mobile networks","authors":"C. Decayeux, D. Semé","doi":"10.1109/ISPDC.2004.4","DOIUrl":"https://doi.org/10.1109/ISPDC.2004.4","url":null,"abstract":"In this work, we propose to extend the standard concept of planar cellular networks into space. Indeed, in cellular networks the trend is to have smaller cells for indoor environment. Then, we give a description of a 3D picocellular network based on layers of hexagonal mesh to ensure that challenge.","PeriodicalId":62714,"journal":{"name":"骈文研究","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2004-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88499751","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：