Pub Date : 1994-01-31DOI: 10.1109/MASCOT.1994.284453
Karim Harzallah, K. Sevcik
The effectiveness of large scale computing depends to a great extent on the performance of the memory system. As shared memory multiprocessors grow in size, their memory hierarchy deepens, resulting in a design with non-uniform latencies. In this paper, we explore the implications of multi-valued memory latencies. In particular, we study the effect of a non-uniform traffic distribution on a hierarchical large scale NUMA multiprocessor named Hector. Memory analysis is of interest because memory is a frequent source of poor performance in large scale multiprocessors. We have developed an analytical model that includes the effects of increased contention for system resources, and the impact of the arbitration algorithm on the network traffic. Our analysis has been validated with a detailed simulator. Also, we have examined two techniques for reducing memory latency. We assess the potential performance gains from replication of data and investigate the improvement in memory utilization by allowing memory request buffering. Furthermore, we studied the sensitivity of the memory performance to changes in background traffic. We found that inter-station traffic has a significant performance effect.<>
{"title":"Evaluating memory system performance of a large scale NUMA multiprocessor","authors":"Karim Harzallah, K. Sevcik","doi":"10.1109/MASCOT.1994.284453","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284453","url":null,"abstract":"The effectiveness of large scale computing depends to a great extent on the performance of the memory system. As shared memory multiprocessors grow in size, their memory hierarchy deepens, resulting in a design with non-uniform latencies. In this paper, we explore the implications of multi-valued memory latencies. In particular, we study the effect of a non-uniform traffic distribution on a hierarchical large scale NUMA multiprocessor named Hector. Memory analysis is of interest because memory is a frequent source of poor performance in large scale multiprocessors. We have developed an analytical model that includes the effects of increased contention for system resources, and the impact of the arbitration algorithm on the network traffic. Our analysis has been validated with a detailed simulator. Also, we have examined two techniques for reducing memory latency. We assess the potential performance gains from replication of data and investigate the improvement in memory utilization by allowing memory request buffering. Furthermore, we studied the sensitivity of the memory performance to changes in background traffic. We found that inter-station traffic has a significant performance effect.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"246 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":"114279144","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.1994.284394
M. Hao, A. Karp, M. Mackey, Vineet Singh, Jane Chien
IVD, Interactive Visualization Debugger is intended to provide on-line and integrated mechanisms for debugging, performance analysis, and data visualization for message-passing parallel applications. The current IVD includes: (I) ESP, a mechanism to multicast window-based commands from a single control window to some subset of existing debuggers/visualizers on various processes; (2) program replay to reproduce program runs deterministically to enable cyclic debugging; and (3) ad hoc data visualization of distributed arrays using existing visualizers.<>
{"title":"On-the-fly visualization and debugging of parallel programs","authors":"M. Hao, A. Karp, M. Mackey, Vineet Singh, Jane Chien","doi":"10.1109/MASCOT.1994.284394","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284394","url":null,"abstract":"IVD, Interactive Visualization Debugger is intended to provide on-line and integrated mechanisms for debugging, performance analysis, and data visualization for message-passing parallel applications. The current IVD includes: (I) ESP, a mechanism to multicast window-based commands from a single control window to some subset of existing debuggers/visualizers on various processes; (2) program replay to reproduce program runs deterministically to enable cyclic debugging; and (3) ad hoc data visualization of distributed arrays using existing visualizers.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"103 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":"130840504","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.1994.284412
Wei Liu, B. Stephens, E. K. Park
An analytical model is developed for performance evaluation of the binary feedback scheme for congestion control in data communication networks. The relationship between the change of window site and the length of the source queue is discovered by introducing an auxiliary virtual packet blocking mechanism. Based on this relationship it is demonstrated that the system throughput can be computed locally by monitoring the traffic in the source node. The model is solved by a new mean valve analysis algorithm. The significance of this study is the provision of a uniform modeling approach which can be generally applied to other feedback control mechanisms. Finally, numerical examples are given to demonstrate the methodology and the validity of the analytical results.<>
{"title":"An analytical model for the binary feedback scheme","authors":"Wei Liu, B. Stephens, E. K. Park","doi":"10.1109/MASCOT.1994.284412","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284412","url":null,"abstract":"An analytical model is developed for performance evaluation of the binary feedback scheme for congestion control in data communication networks. The relationship between the change of window site and the length of the source queue is discovered by introducing an auxiliary virtual packet blocking mechanism. Based on this relationship it is demonstrated that the system throughput can be computed locally by monitoring the traffic in the source node. The model is solved by a new mean valve analysis algorithm. The significance of this study is the provision of a uniform modeling approach which can be generally applied to other feedback control mechanisms. Finally, numerical examples are given to demonstrate the methodology and the validity of the analytical results.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"61 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":"126710852","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.1994.284415
E. Gelenbe, R. Kushwaha
We design and evaluate novel adaptive load balancing algorithms for processes in an MIMD architecture. We introduce a gradient descent paradigm to compute on-line load balancing decisions and implement it on an nCube target architecture. Our measurements show that such algorithms will substantially improve distributed system performance.<>
{"title":"Dynamic load balancing in distributed systems","authors":"E. Gelenbe, R. Kushwaha","doi":"10.1109/MASCOT.1994.284415","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284415","url":null,"abstract":"We design and evaluate novel adaptive load balancing algorithms for processes in an MIMD architecture. We introduce a gradient descent paradigm to compute on-line load balancing decisions and implement it on an nCube target architecture. Our measurements show that such algorithms will substantially improve distributed system performance.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"16 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":"114821262","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.1994.284446
P. M. Greenawalt
Recently, a large effort has been made to reduce the power consumed by computer-systems. Multiple power states have been defined, and mechanisms have been developed to allow system software to control transitions between these states. Unfortunately, little work has been done to determine effective times to change states. Statistical models of the power utilized by individual subsystems call provide a basis for making such decisions. The hard disk state model described provides both the quantitative data and insight necessary to design an efficient power management system.<>
{"title":"Modeling power management for hard disks","authors":"P. M. Greenawalt","doi":"10.1109/MASCOT.1994.284446","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284446","url":null,"abstract":"Recently, a large effort has been made to reduce the power consumed by computer-systems. Multiple power states have been defined, and mechanisms have been developed to allow system software to control transitions between these states. Unfortunately, little work has been done to determine effective times to change states. Statistical models of the power utilized by individual subsystems call provide a basis for making such decisions. The hard disk state model described provides both the quantitative data and insight necessary to design an efficient power management system.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"61 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":"128446470","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.1994.284428
J. B. Sinclair, W. P. Dawkins
ES is a tool for estimating the execution times of parallel algorithms on MIMD parallel systems. ES allows the user to model arbitrary task execution times, explicit task precedence and synchronization constraints, resource contention among tasks, and a variety of scheduling policies for shared resources. Given a model of a parallel algorithm and a parallel system, ES constructs a sequencing tree that represents some or all of the possible sequences of events that may occur during the execution of the algorithm on the system, and uses it to estimate the mean and standard deviation of the execution time of the parallel algorithm. The authors compare estimates generated by ES to measurements made of a parallel mergesort executing on an Intel iPSC/860 hypercube.<>
{"title":"ES: a tool for predicting the performance of parallel systems","authors":"J. B. Sinclair, W. P. Dawkins","doi":"10.1109/MASCOT.1994.284428","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284428","url":null,"abstract":"ES is a tool for estimating the execution times of parallel algorithms on MIMD parallel systems. ES allows the user to model arbitrary task execution times, explicit task precedence and synchronization constraints, resource contention among tasks, and a variety of scheduling policies for shared resources. Given a model of a parallel algorithm and a parallel system, ES constructs a sequencing tree that represents some or all of the possible sequences of events that may occur during the execution of the algorithm on the system, and uses it to estimate the mean and standard deviation of the execution time of the parallel algorithm. The authors compare estimates generated by ES to measurements made of a parallel mergesort executing on an Intel iPSC/860 hypercube.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"50 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":"130886237","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.1994.284436
W. Najjar, A. Lagman, S. Sur, P. Srimani
Interest in parallel processing has steadily increased in recent years, and correspondingly, modeling techniques for communication systems have become increasingly important. In this paper we present a performance evaluation study of adaptive routing strategies in k-ary n-cube interconnection networks. Adaptive routing, as opposed to deterministic routing, exploits the presence of multiple paths in a network in order to reduce communication latency. Two minimal path adaptive strategies are evaluated using extensive simulation experiments. In addition, we derive an analytic model using Markov queueing for the case when n=2. We use these models to measure the communication speedup that can be achieved by adaptive routing, and to assess the impact of various network parameters on the performance of such systems.<>
{"title":"Modeling adaptive routing in k-ary n-cube networks","authors":"W. Najjar, A. Lagman, S. Sur, P. Srimani","doi":"10.1109/MASCOT.1994.284436","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284436","url":null,"abstract":"Interest in parallel processing has steadily increased in recent years, and correspondingly, modeling techniques for communication systems have become increasingly important. In this paper we present a performance evaluation study of adaptive routing strategies in k-ary n-cube interconnection networks. Adaptive routing, as opposed to deterministic routing, exploits the presence of multiple paths in a network in order to reduce communication latency. Two minimal path adaptive strategies are evaluated using extensive simulation experiments. In addition, we derive an analytic model using Markov queueing for the case when n=2. We use these models to measure the communication speedup that can be achieved by adaptive routing, and to assess the impact of various network parameters on the performance of such systems.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"6 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":"124503250","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.1994.284393
Li-Tao Chen, G. Memmi, Pascal Petit, Patrick Denimal
Distributed collaboration systems intend to provide software engineers as well as end-users the facilities to easily control and process their collaborative activities. This paper investigates our Scrutiny/CB architecture and environment through real usages and extensive simulation experiments. The analysis and evaluation results from this study provide insight observations and helpful guidance to the future improvement of distributed collaboration systems.<>
{"title":"Experiment and performance evaluation of a distributed collaboration system","authors":"Li-Tao Chen, G. Memmi, Pascal Petit, Patrick Denimal","doi":"10.1109/MASCOT.1994.284393","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284393","url":null,"abstract":"Distributed collaboration systems intend to provide software engineers as well as end-users the facilities to easily control and process their collaborative activities. This paper investigates our Scrutiny/CB architecture and environment through real usages and extensive simulation experiments. The analysis and evaluation results from this study provide insight observations and helpful guidance to the future improvement of distributed collaboration systems.<<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":"130458216","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.1994.284417
D. Chao, David T. Wang
The Petri net (PN) can be incrementally expanded by adding paths of nodes according to the synthesis rules. These paths of nodes are generated to increase the number of concurrent, conditional, and iterative processes. A set of synthesis rules has been developed to ensure the correct generation of these paths of nodes. The advantages of these rules include the possibility of implementation as a CAD tool, the guarantee of liveness, boundedness, and reversibility (well-behaved properties), and the close control of concurrency and nondeterminism. These rules serve two functions for interactive design: prevention of potential problems and guidance of correct new path generations. Examples of synthesis and a procedure of finding invariants are presented to show that one can synthesize PNs beyond asymmetric-choice nets. The technique is applied to the synthesis of communication protocols and extended to synthesize local entities.<>
{"title":"The knitting technique and its application to communication protocol synthesis","authors":"D. Chao, David T. Wang","doi":"10.1109/MASCOT.1994.284417","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284417","url":null,"abstract":"The Petri net (PN) can be incrementally expanded by adding paths of nodes according to the synthesis rules. These paths of nodes are generated to increase the number of concurrent, conditional, and iterative processes. A set of synthesis rules has been developed to ensure the correct generation of these paths of nodes. The advantages of these rules include the possibility of implementation as a CAD tool, the guarantee of liveness, boundedness, and reversibility (well-behaved properties), and the close control of concurrency and nondeterminism. These rules serve two functions for interactive design: prevention of potential problems and guidance of correct new path generations. Examples of synthesis and a procedure of finding invariants are presented to show that one can synthesize PNs beyond asymmetric-choice nets. The technique is applied to the synthesis of communication protocols and extended to synthesize local entities.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"45 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":"134451402","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.1994.284388
Bin Zhou, Mohammed Atiquzzaman
The objective of the authors is to develop an analytical model to evaluate the throughput and packet delay, in the presence of nonuniform traffic, in a MIN having multiple buffers at the output of the switching elements. The model is based on Markov chains, and uses several simplifying assumptions to make the model tractable. The validity of the model has been verified by comparison of the results with those obtained from simulations.<>
{"title":"Performance of output-multibuffered multistage interconnection networks under non-uniform traffic patterns","authors":"Bin Zhou, Mohammed Atiquzzaman","doi":"10.1109/MASCOT.1994.284388","DOIUrl":"https://doi.org/10.1109/MASCOT.1994.284388","url":null,"abstract":"The objective of the authors is to develop an analytical model to evaluate the throughput and packet delay, in the presence of nonuniform traffic, in a MIN having multiple buffers at the output of the switching elements. The model is based on Markov chains, and uses several simplifying assumptions to make the model tractable. The validity of the model has been verified by comparison of the results with those obtained from simulations.<<ETX>>","PeriodicalId":288344,"journal":{"name":"Proceedings of International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems","volume":"53 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":"132382735","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：