Both experimental and theoretical approaches to high energy physics (HEP) at Fermilab are computer technology limited: no one can identify a requirement on computing or data capacity that is independent of cost or other realities. Fermilab has been forced, therefore, to turn significant attention and resources to finding extremely cost effective solutions to its computing using whatever technology is available, and in the process has become a pioneer in goal-driven computer science, integrating commercial solution at the chip, board, and system level. In recent years, most obviously first in the hypercube movement, and now in the surge of interest in networked parallel workstation clusters,'' explicit parallelism has become a recognized force in the computer science attack on the problem of parallel computing. The report discusses the implementation of these computer systems.
{"title":"High Performance Parallel Local Memory Computing at Fermilab","authors":"T. Nash","doi":"10.1109/WHP.1992.664392","DOIUrl":"https://doi.org/10.1109/WHP.1992.664392","url":null,"abstract":"Both experimental and theoretical approaches to high energy physics (HEP) at Fermilab are computer technology limited: no one can identify a requirement on computing or data capacity that is independent of cost or other realities. Fermilab has been forced, therefore, to turn significant attention and resources to finding extremely cost effective solutions to its computing using whatever technology is available, and in the process has become a pioneer in goal-driven computer science, integrating commercial solution at the chip, board, and system level. In recent years, most obviously first in the hypercube movement, and now in the surge of interest in networked parallel workstation clusters,'' explicit parallelism has become a recognized force in the computer science attack on the problem of parallel computing. The report discusses the implementation of these computer systems.","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125565340","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}
Most scientific applications are currently structured as a series of computational steps, each of which is implemented by a separate program with jiles being used to transmit data between the steps. Here, an alternative model is described in which the application is constructed as a heterogeneous, distributed program in order to improve facilities for user interaction. The specifrc focus is on describing how AVS (Application Visualization System) from AVS, Inc. has been adapted to support this type of application using the Schooner RPC system. An example involving self-organizing neural nets is used to illustrate the details of this approach.
{"title":"Adapting AVS to Support Scientific Applications As Heterogeneous, Distributed Programs","authors":"Patrick T. Homer, R. Schlichting","doi":"10.1109/WHP.1992.664385","DOIUrl":"https://doi.org/10.1109/WHP.1992.664385","url":null,"abstract":"Most scientific applications are currently structured as a series of computational steps, each of which is implemented by a separate program with jiles being used to transmit data between the steps. Here, an alternative model is described in which the application is constructed as a heterogeneous, distributed program in order to improve facilities for user interaction. The specifrc focus is on describing how AVS (Application Visualization System) from AVS, Inc. has been adapted to support this type of application using the Schooner RPC system. An example involving self-organizing neural nets is used to illustrate the details of this approach.","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"138 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124205318","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}
Large meta-systems comprised of a variety of interconnected high-performance architectures are becoming available to researchers. To fully exploit these new systems software support must be provided that is easy to use, supports large degrees of parallelism in applications code, and manages the complexity of the underlying physical architecture for the user, This paper describes our approach to constructing and exploiting meta-systems. Our approach inherits features of earlier work on parallel processing systems and heterogeneous distributed computing systems. In particular, we build on Mentat, an object-oriented parallel processing system developed at the University of Virginia that provides large amounts of easy-to-use parallelism for MIMD architectures.'
{"title":"Meta-Systems: An Approach Combining Parallel Processing and Heterogeneous Distributed Computing Systems","authors":"A. Grimshaw","doi":"10.1109/WHP.1992.664386","DOIUrl":"https://doi.org/10.1109/WHP.1992.664386","url":null,"abstract":"Large meta-systems comprised of a variety of interconnected high-performance architectures are becoming available to researchers. To fully exploit these new systems software support must be provided that is easy to use, supports large degrees of parallelism in applications code, and manages the complexity of the underlying physical architecture for the user, This paper describes our approach to constructing and exploiting meta-systems. Our approach inherits features of earlier work on parallel processing systems and heterogeneous distributed computing systems. In particular, we build on Mentat, an object-oriented parallel processing system developed at the University of Virginia that provides large amounts of easy-to-use parallelism for MIMD architectures.'","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"74 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114289630","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 paper, we study an approach called superconcurrency for identifying and employing an optimal, heterogeneous suite of processors to solve traditional supercomputing problems. We discuss mainly two components of superconcurrency: Optimal selection theory, and Parallel programming tools. We briefly describe a number of existing parallel programming tools and present a parallel programming model called Cluster-M. This model efficiently provides a programming paradigm for high-order heterogeneous procedural specification computing. Using ClusterM, portable software can be generated which may be mapped onto various configurations of heterogeneous supercomputing.
{"title":"Cluster-M Paradigms for High-Order Heterogeneous Procedural Specification Computing","authors":"M. M. Eshaghian, R. F. Freund","doi":"10.1109/WHP.1992.664384","DOIUrl":"https://doi.org/10.1109/WHP.1992.664384","url":null,"abstract":"In this paper, we study an approach called superconcurrency for identifying and employing an optimal, heterogeneous suite of processors to solve traditional supercomputing problems. We discuss mainly two components of superconcurrency: Optimal selection theory, and Parallel programming tools. We briefly describe a number of existing parallel programming tools and present a parallel programming model called Cluster-M. This model efficiently provides a programming paradigm for high-order heterogeneous procedural specification computing. Using ClusterM, portable software can be generated which may be mapped onto various configurations of heterogeneous supercomputing.","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133723695","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}
This paper analyzes the coexistence of heterogeneity and multiprogramming an parallel systems. Using simulation models, we compare the performance of cost equivalent homogeneous and heterogeneous multiprogrammed parallel systems. In particular, we find that heterogeneity reduces the average response time an a multiprogrammed system. We also investigate new processor schedukng policies that take advantage of heterogeneity to improve response time in a multiprogrammed e nvironm ent .
{"title":"The Effect of Heterogeneity on the Performance of Multiprogrammed Parallel Systems","authors":"Virgílio A. F. Almeida, I. Vasconcelos, J. Árabe","doi":"10.1109/WHP.1992.664381","DOIUrl":"https://doi.org/10.1109/WHP.1992.664381","url":null,"abstract":"This paper analyzes the coexistence of heterogeneity and multiprogramming an parallel systems. Using simulation models, we compare the performance of cost equivalent homogeneous and heterogeneous multiprogrammed parallel systems. In particular, we find that heterogeneity reduces the average response time an a multiprogrammed system. We also investigate new processor schedukng policies that take advantage of heterogeneity to improve response time in a multiprogrammed e nvironm ent .","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"77 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126582674","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}
We present a technique for partitioning signal processing algorithms represented as signal flow graphs for execution on a hierarchically structured, heterogeneous multiprocessor architecture. We propose the use of an optimization algorithm with an uduptive seurch schedule to find good partitions. We find that such an approach consistently determines optimal partitions for several example signal flow graphs. The paper discusses the present form of the algorithm, and results of an experimental evaluation.
{"title":"Partitioning Signal Flow Graphs for Execution on Heterogeneous Signal Processing Architectures","authors":"C. De Castro, S. Yalamanchili","doi":"10.1109/WHP.1992.664389","DOIUrl":"https://doi.org/10.1109/WHP.1992.664389","url":null,"abstract":"We present a technique for partitioning signal processing algorithms represented as signal flow graphs for execution on a hierarchically structured, heterogeneous multiprocessor architecture. We propose the use of an optimization algorithm with an uduptive seurch schedule to find good partitions. We find that such an approach consistently determines optimal partitions for several example signal flow graphs. The paper discusses the present form of the algorithm, and results of an experimental evaluation.","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"29 6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125695190","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 increasing demand for grealer performance along with recent advances an high performance networking have transformed the traditional supercomputer model of a large central machine into a distributed, networked model of computing where shared terabyte storage systems, high speed graphical frame buflers, high resolution workstations, vector supercomputers and massively parallel computers can be applied together in a cooperative network in order to solve scientific problems of unprecedented complexity. In this paper we describe the High Performance Parallel Interface (HIPPI) operating environment currently in place at the Minnesota Supercomputer Center and provide preliminary performance measurements obtained while transmitting data between a CRAY-2 and CM-2 computer system. Finally, we discuss how the HIPPI can be used by compute-intensive applications to reduce their overall execution time.
{"title":"Network Supercomputing: Experiments with a Cray-2 to Cm-2 Hippi Connection","authors":"R. Vetter, D. Du, A. Klietz","doi":"10.1109/WHP.1992.664390","DOIUrl":"https://doi.org/10.1109/WHP.1992.664390","url":null,"abstract":"The increasing demand for grealer performance along with recent advances an high performance networking have transformed the traditional supercomputer model of a large central machine into a distributed, networked model of computing where shared terabyte storage systems, high speed graphical frame buflers, high resolution workstations, vector supercomputers and massively parallel computers can be applied together in a cooperative network in order to solve scientific problems of unprecedented complexity. In this paper we describe the High Performance Parallel Interface (HIPPI) operating environment currently in place at the Minnesota Supercomputer Center and provide preliminary performance measurements obtained while transmitting data between a CRAY-2 and CM-2 computer system. Finally, we discuss how the HIPPI can be used by compute-intensive applications to reduce their overall execution time.","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130973447","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}
This paper discusses a framework for supporting heterogeneous computing systems (HCS). HCS are physically distributed, potentially complex and evolving systems. HCS consist of a number of computers and interconnections which may have differing language and system support, and sometimes distinct computational (architectural) models. By providing a large number of computational resources to the user, HCS have the potential to allow highly eficient execution of ultra largescale applications. The paper provides a framework for addressing some of the significant problems in interoperability and resource management which result from the heterogeneity in HCS. It represents ideas and ongoing research in the area.
{"title":"An Actor-Based Framework for Heterogeneous Computing Systems","authors":"Gul A. Agha, R. Panwar","doi":"10.1109/WHP.1992.664382","DOIUrl":"https://doi.org/10.1109/WHP.1992.664382","url":null,"abstract":"This paper discusses a framework for supporting heterogeneous computing systems (HCS). HCS are physically distributed, potentially complex and evolving systems. HCS consist of a number of computers and interconnections which may have differing language and system support, and sometimes distinct computational (architectural) models. By providing a large number of computational resources to the user, HCS have the potential to allow highly eficient execution of ultra largescale applications. The paper provides a framework for addressing some of the significant problems in interoperability and resource management which result from the heterogeneity in HCS. It represents ideas and ongoing research in the area.","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127425210","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}
{"title":"Linda in Heterogeneous Computing Environments","authors":"N. Carriero, D. Gelernter, T. Mattson","doi":"10.1109/WHP.1992.664383","DOIUrl":"https://doi.org/10.1109/WHP.1992.664383","url":null,"abstract":"","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124931832","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}
We discuss the development and implementation of a component of a heterogeneous supercomputing environment at the Pittsburgh Supercomputing Center (PSC). This component, built upon a CRAY Y-W/832, a Thinking Machines CM-2, and HiPPI communications channel, allows applications to be partitioned between the two supercomputers. Development work included the design of a user environment and performance analysis of the resulting system. Implementation involved tackling communications and data presentation issues in the heterogeneous environment. This work provides a case study of a heterogeneous supercomputing environment, showing strengths and weaknesses of currently available technology along with directions for future development.
{"title":"Deployment of a Hippi-Based Distributed Supercomputing Environment at the Pittsburgh Supercomputing Center","authors":"J. Mahdavi, G.L. Huntoon, M. Mathis","doi":"10.1109/WHP.1992.664391","DOIUrl":"https://doi.org/10.1109/WHP.1992.664391","url":null,"abstract":"We discuss the development and implementation of a component of a heterogeneous supercomputing environment at the Pittsburgh Supercomputing Center (PSC). This component, built upon a CRAY Y-W/832, a Thinking Machines CM-2, and HiPPI communications channel, allows applications to be partitioned between the two supercomputers. Development work included the design of a user environment and performance analysis of the resulting system. Implementation involved tackling communications and data presentation issues in the heterogeneous environment. This work provides a case study of a heterogeneous supercomputing environment, showing strengths and weaknesses of currently available technology along with directions for future development.","PeriodicalId":201815,"journal":{"name":"Proceedings. Workshop on Heterogeneous Processing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122222073","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：