Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284434
R. Thakur, A. Choudhary, G. Fox
The complete exchange (or all-to-all personalized) communication pattern occurs frequently in many important parallel computing applications. We discuss several algorithms to perform complete exchange on a two dimensional mesh connected computer with wormhole routing. We propose algorithms for both power-of-two and non power-of-two meshes as well as an algorithm which works for any arbitrary mesh. We have developed analytical models to estimate the performance of the algorithms on the basis of system parameters. These models take into account the effects of link contention and other characteristics of the communication system. Performance results on the Intel Touchstone Delta are presented and analyzed.<>
{"title":"Complete exchange on a wormhole routed mesh","authors":"R. Thakur, A. Choudhary, G. Fox","doi":"10.1109/MASCOT.1994.284434","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284434","url":null,"abstract":"The complete exchange (or all-to-all personalized) communication pattern occurs frequently in many important parallel computing applications. We discuss several algorithms to perform complete exchange on a two dimensional mesh connected computer with wormhole routing. We propose algorithms for both power-of-two and non power-of-two meshes as well as an algorithm which works for any arbitrary mesh. We have developed analytical models to estimate the performance of the algorithms on the basis of system parameters. These models take into account the effects of link contention and other characteristics of the communication system. Performance results on the Intel Touchstone Delta are presented and analyzed.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"107 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123246468","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284422
J. Veenstra, R. Fowler
MINT is a software package designed to ease the process of constructing event-driven memory hierarchy simulators for multiprocessors. It provides a set of simulated processors that run standard Unix executable files compiled for a MIPS R3000 based multiprocessor. These generate multiple streams of memory reference events that drive a user-provided memory system simulator. MINT uses a novel hybrid technique that exploits the best aspects of native execution and software interpretation to minimize the overhead of processor simulation. Combined with related techniques to improve performance, this approach makes simulation on uniprocessor hosts extremely efficient.<>
{"title":"MINT: a front end for efficient simulation of shared-memory multiprocessors","authors":"J. Veenstra, R. Fowler","doi":"10.1109/MASCOT.1994.284422","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284422","url":null,"abstract":"MINT is a software package designed to ease the process of constructing event-driven memory hierarchy simulators for multiprocessors. It provides a set of simulated processors that run standard Unix executable files compiled for a MIPS R3000 based multiprocessor. These generate multiple streams of memory reference events that drive a user-provided memory system simulator. MINT uses a novel hybrid technique that exploits the best aspects of native execution and software interpretation to minimize the overhead of processor simulation. Combined with related techniques to improve performance, this approach makes simulation on uniprocessor hosts extremely efficient.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"418 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115996200","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284457
R. Puigjaner, J. Szymanski
Real-time and embedded systems always have performance constraints. With conventional design techniques, these constraints are verified at the testing phase and the designer does no performance constraint estimation during the design. From the design information, it is possible to deduce a queueing network performance model, but its execution requires supplementary information concerning estimations of the execution times. Delays generated by logical locking and by resource use conflicts can be computed, via analytical or simulation techniques, from such a model. This paper presents experience acquired in the integration of design methods and the performance modelling techniques.<>
{"title":"Towards the automatic derivation of computer performance models from the real time and embedded systems design","authors":"R. Puigjaner, J. Szymanski","doi":"10.1109/MASCOT.1994.284457","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284457","url":null,"abstract":"Real-time and embedded systems always have performance constraints. With conventional design techniques, these constraints are verified at the testing phase and the designer does no performance constraint estimation during the design. From the design information, it is possible to deduce a queueing network performance model, but its execution requires supplementary information concerning estimations of the execution times. Delays generated by logical locking and by resource use conflicts can be computed, via analytical or simulation techniques, from such a model. This paper presents experience acquired in the integration of design methods and the performance modelling techniques.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121899072","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284406
Cedell Alexander, D. Reese, J. Harden
Program activity graphs can be constructed from time-stamped traces of appropriate execution events. Information about the activities on the k longest execution paths is useful in the analysis of parallel program performance. In this paper, four algorithms for finding the near-critical paths of program activity graphs are presented and compared, including an efficient new algorithm that utilizes slack values calculated by the critical path method to perform a best-first search in linear space. The worst-case time and memory requirements of the new algorithm are in O(ke) and O(k+e), where e is the number of edges in the graph. Results confirming the efficiency of the algorithm are presented for five application programs. A framework for utilizing the near-critical path information is also described. The framework includes both statistical summaries and visualization capabilities.<>
{"title":"Near-critical path analysis of program activity graphs","authors":"Cedell Alexander, D. Reese, J. Harden","doi":"10.1109/MASCOT.1994.284406","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284406","url":null,"abstract":"Program activity graphs can be constructed from time-stamped traces of appropriate execution events. Information about the activities on the k longest execution paths is useful in the analysis of parallel program performance. In this paper, four algorithms for finding the near-critical paths of program activity graphs are presented and compared, including an efficient new algorithm that utilizes slack values calculated by the critical path method to perform a best-first search in linear space. The worst-case time and memory requirements of the new algorithm are in O(ke) and O(k+e), where e is the number of edges in the graph. Results confirming the efficiency of the algorithm are presented for five application programs. A framework for utilizing the near-critical path information is also described. The framework includes both statistical summaries and visualization capabilities.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116880893","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284413
W. Stewart, A. Touzene
Iterative aggregation/disaggregation (IAD) methods are powerful tools for solving Markov chain models whose transition probability matrices are nearly completely decomposable (NCD). Such models arise frequently during the performance and reliability analysis of computer and telecommunication systems. IAD methods require the solution of a stochastic coupling matrix whose elements denote transition probabilities among blocks. The coupling matrices are often large and in NCD models necessarily have diagonal elements close to one and small off-diagonal elements. This makes their solution by either iterative or direct methods rather difficult. We propose a modification of the coupling matrix that allows us to accurate and efficiently compute its stationary probability vector.<>
{"title":"On solving stochastic coupling matrices arising in iterative aggregation/disaggregation methods","authors":"W. Stewart, A. Touzene","doi":"10.1109/MASCOT.1994.284413","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284413","url":null,"abstract":"Iterative aggregation/disaggregation (IAD) methods are powerful tools for solving Markov chain models whose transition probability matrices are nearly completely decomposable (NCD). Such models arise frequently during the performance and reliability analysis of computer and telecommunication systems. IAD methods require the solution of a stochastic coupling matrix whose elements denote transition probabilities among blocks. The coupling matrices are often large and in NCD models necessarily have diagonal elements close to one and small off-diagonal elements. This makes their solution by either iterative or direct methods rather difficult. We propose a modification of the coupling matrix that allows us to accurate and efficiently compute its stationary probability vector.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"78 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124850113","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284419
Marc Beckers, J. Peeters, F. Verboven
Development of a network management system for B-ISDN access products proceeded through a modelling phase at a high level of abstraction, followed by implementation in an expert-system shell dedicated to network management. The expert-system shell provides the opportunity to exercise the system in simulation mode. The object-oriented paradigm is used for modelling, implementation and simulation. Some of the modelling techniques employed are highlighted and followed by an overview of the environment which simulates and implements the system. Finally, some strategies for transforming a model at a high level of abstraction into an implementation are discussed.<>
{"title":"Object-oriented modeling, simulation and implementation of a network management system","authors":"Marc Beckers, J. Peeters, F. Verboven","doi":"10.1109/MASCOT.1994.284419","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284419","url":null,"abstract":"Development of a network management system for B-ISDN access products proceeded through a modelling phase at a high level of abstraction, followed by implementation in an expert-system shell dedicated to network management. The expert-system shell provides the opportunity to exercise the system in simulation mode. The object-oriented paradigm is used for modelling, implementation and simulation. Some of the modelling techniques employed are highlighted and followed by an overview of the environment which simulates and implements the system. Finally, some strategies for transforming a model at a high level of abstraction into an implementation are discussed.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124866381","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284398
R. Trieb, E. Puttkamer
Presents the 3d7-Simulation Environment, which was designed as a tool for wheeled autonomous mobile robot development. Within the MOBOT-IV project 3d7 is used to support the research in the field of sensor data processing algorithms as well as the different levels of motion control. Currently, the simulation environment consists of a 3D solid modeling kernel, environment editor, optical range finder sensor simulator, sonar sensor simulation component and kinematics and dynamics simulator for wheeled vehicles.<>
{"title":"The 3d7-Simulation Environment: a tool for autonomous mobile robot development","authors":"R. Trieb, E. Puttkamer","doi":"10.1109/MASCOT.1994.284398","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284398","url":null,"abstract":"Presents the 3d7-Simulation Environment, which was designed as a tool for wheeled autonomous mobile robot development. Within the MOBOT-IV project 3d7 is used to support the research in the field of sensor data processing algorithms as well as the different levels of motion control. Currently, the simulation environment consists of a 3D solid modeling kernel, environment editor, optical range finder sensor simulator, sonar sensor simulation component and kinematics and dynamics simulator for wheeled vehicles.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129701306","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284424
P. Mehra, M. Gower, Michael A. Bass
We present a system for automated modeling of message-passing programs. Its models preserve the parallel program's structure, especially the syntactic boundaries surrounding communication calls. Our grammar-driven approach uses the program's parse trees to derive a regular expression that describes all possible execution traces at the chosen level of modeling; that expression is used for automatic extraction of timing information from traces of scaled-down runs. We consider "intelligent regression" techniques for discovering the numerical attributes of our models: run times of sequential blocks; lengths and destinations of messages; and loop bounds. Regression produces formulae expressing these attributes in terms of problem and system sizes. The model is then used for predicting the performance of large-scale runs. We illustrate our approach with a program that simultaneously solves multiple tridiagonal linear systems an the iPSC/860.<>
{"title":"Automated modeling of message-passing programs","authors":"P. Mehra, M. Gower, Michael A. Bass","doi":"10.1109/MASCOT.1994.284424","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284424","url":null,"abstract":"We present a system for automated modeling of message-passing programs. Its models preserve the parallel program's structure, especially the syntactic boundaries surrounding communication calls. Our grammar-driven approach uses the program's parse trees to derive a regular expression that describes all possible execution traces at the chosen level of modeling; that expression is used for automatic extraction of timing information from traces of scaled-down runs. We consider \"intelligent regression\" techniques for discovering the numerical attributes of our models: run times of sequential blocks; lengths and destinations of messages; and loop bounds. Regression produces formulae expressing these attributes in terms of problem and system sizes. The model is then used for predicting the performance of large-scale runs. We illustrate our approach with a program that simultaneously solves multiple tridiagonal linear systems an the iPSC/860.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125421479","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284421
Gudjon Hermannsson, Ai Li, L. Wittie
This paper introduces the EC frontend and DSIM simulator. Given a parallel program, they determine its execution time on huge networks of computers. EC extracts task step needs. DSIM predicts completion times rather than simulating each program step. This paper contains analyses of the memory savings and the execution time savings for simulations of one to 2,800 computers running parallel Gaussian elimination and fast Fourier transform. The time savings are 20% (two days) for fifty runs of Gaussian reduction of a 400x401 matrix to solve 400 linear equations. Memory needs are reduced 99% (637 MBytes) per simulation run. The memory savings allow simulation of parallel programs running on thousands of processors. These huge network sizes are impractical with step-by-step simulations.<>
{"title":"Analysis of memory and time savings using EC/DSIM","authors":"Gudjon Hermannsson, Ai Li, L. Wittie","doi":"10.1109/MASCOT.1994.284421","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284421","url":null,"abstract":"This paper introduces the EC frontend and DSIM simulator. Given a parallel program, they determine its execution time on huge networks of computers. EC extracts task step needs. DSIM predicts completion times rather than simulating each program step. This paper contains analyses of the memory savings and the execution time savings for simulations of one to 2,800 computers running parallel Gaussian elimination and fast Fourier transform. The time savings are 20% (two days) for fifty runs of Gaussian reduction of a 400x401 matrix to solve 400 linear equations. Memory needs are reduced 99% (637 MBytes) per simulation run. The memory savings allow simulation of parallel programs running on thousands of processors. These huge network sizes are impractical with step-by-step simulations.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"80 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131885503","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}
Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284443
K. Sivalingam, P. Dowd
Network topology significantly affects the system performance of local area networks. The bus topology was considered for optical local area networks since the degradation of system performance due to propagation delay could be reduced. Power budget constraints severely limited maximum system size which diminished its appeal. On the other hand, the star topology was shown to have excellent fanout and fault tolerance that led to extensive research interest in star based systems. Recent advances in optical amplifiers which expand the power budget have renewed interest in optical bus based networks. The objective of the authors is to compare the performance of bus and star based systems. They assess the performance improvement of bus based approaches, and show under what conditions it is justified in terms of performance, relative to the increase in component costs.<>
{"title":"A performance study of photonic local area network topologies","authors":"K. Sivalingam, P. Dowd","doi":"10.1109/MASCOT.1994.284443","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284443","url":null,"abstract":"Network topology significantly affects the system performance of local area networks. The bus topology was considered for optical local area networks since the degradation of system performance due to propagation delay could be reduced. Power budget constraints severely limited maximum system size which diminished its appeal. On the other hand, the star topology was shown to have excellent fanout and fault tolerance that led to extensive research interest in star based systems. Recent advances in optical amplifiers which expand the power budget have renewed interest in optical bus based networks. The objective of the authors is to compare the performance of bus and star based systems. They assess the performance improvement of bus based approaches, and show under what conditions it is justified in terms of performance, relative to the increase in component costs.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133609524","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：