Pub Date : 1991-04-28DOI: 10.1109/DMCC.1991.633113
R. Shankar, S. Ranka
This paper describes parallel algorithms for the following operations on qua.dtreesboolean operations (union, intersection, complement), collapsing a quadtree, and neighbor finding in an image represented by a quadtree. The architecture assumed in this paper is a. hypercube with one processing element (PE) per hypercube node. We assume that the architecture is SIMD, i.e., all PEs work under the control of a single control unit. [( eywords: Qua.dtrees, Hypercube algorithms, Image processing
{"title":"Hypercube Algorithms for Operations on Quadtrees","authors":"R. Shankar, S. Ranka","doi":"10.1109/DMCC.1991.633113","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633113","url":null,"abstract":"This paper describes parallel algorithms for the following operations on qua.dtreesboolean operations (union, intersection, complement), collapsing a quadtree, and neighbor finding in an image represented by a quadtree. The architecture assumed in this paper is a. hypercube with one processing element (PE) per hypercube node. We assume that the architecture is SIMD, i.e., all PEs work under the control of a single control unit. [( eywords: Qua.dtrees, Hypercube algorithms, Image processing","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"144 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114276413","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633144
S. Seidel, Ming-Homg Lee, S. Fotedar
Each node of the Intel iPSCI2 and iPSCl860 is capable of concurrent bidirectional communication, that is, each node can concurrently send and receive a message. However, link contention created by the long message communication protocol usually interferes with this source of concurrency so that half of the total capacity of the communication network is wasted. This intelference can be overcome by synchronizing communicating pairs of nodes. The cost of algorith,ms for the simple exchange, the one-to-all broadcast, the allto-all broadcast, and the complete exchange operations can be significantly reduced, compared to the cost of their usual implementations, by exploiting this approach. Performance measurements of algorithms for these communication problems support these observations.
{"title":"Concurrent Bidirectional Communication On The Intel iPSC/860 And iPSC/2","authors":"S. Seidel, Ming-Homg Lee, S. Fotedar","doi":"10.1109/DMCC.1991.633144","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633144","url":null,"abstract":"Each node of the Intel iPSCI2 and iPSCl860 is capable of concurrent bidirectional communication, that is, each node can concurrently send and receive a message. However, link contention created by the long message communication protocol usually interferes with this source of concurrency so that half of the total capacity of the communication network is wasted. This intelference can be overcome by synchronizing communicating pairs of nodes. The cost of algorith,ms for the simple exchange, the one-to-all broadcast, the allto-all broadcast, and the complete exchange operations can be significantly reduced, compared to the cost of their usual implementations, by exploiting this approach. Performance measurements of algorithms for these communication problems support these observations.","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"2012 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114749509","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633093
L. Welch
Software reuse increases the productivity of software engineers, however, the benefits of reuse can be offs'et by poor performance of the softwane products. To avoid this pitfall, the potential inefficiencies of a class of reusable software components are identified and addressed. An architecture providing a model of parallel ezecution that not only speeds up programs, but enc,ourages the development of layered software by increasing parallelism in correspondence to increases in layering, is developed. The model encourages development of highly cohesive, loosely coupled' modules. To improve the effectiveness of the model of parallel e.zecution, static mapping techniques are used in conjunction with dynamic scheduling and dynamic data distribution techniques. The combination of static and dynamic load balancing techniques allows the benefits of each technique to be ezploited while reducing its negative effects. To avoid the expense of dynamically neplicating (cloning) code, or of movin:g code between PES, modules are statically cloned and mapped to PES. Calls are scheduled dynamically and doita are mapped dynamically, since it is impossible to determine before run-time the execution sequence of a program, and bec(o.use these techniques introduce very little overhead at run-time (due to simplicity and hardware implementation). Additionally, the techniques used for scheduling and data distribution are distributed to avoid serialization.
{"title":"Architectural Support for Dynamic Data Distribution and Dynamic","authors":"L. Welch","doi":"10.1109/DMCC.1991.633093","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633093","url":null,"abstract":"Software reuse increases the productivity of software engineers, however, the benefits of reuse can be offs'et by poor performance of the softwane products. To avoid this pitfall, the potential inefficiencies of a class of reusable software components are identified and addressed. An architecture providing a model of parallel ezecution that not only speeds up programs, but enc,ourages the development of layered software by increasing parallelism in correspondence to increases in layering, is developed. The model encourages development of highly cohesive, loosely coupled' modules. To improve the effectiveness of the model of parallel e.zecution, static mapping techniques are used in conjunction with dynamic scheduling and dynamic data distribution techniques. The combination of static and dynamic load balancing techniques allows the benefits of each technique to be ezploited while reducing its negative effects. To avoid the expense of dynamically neplicating (cloning) code, or of movin:g code between PES, modules are statically cloned and mapped to PES. Calls are scheduled dynamically and doita are mapped dynamically, since it is impossible to determine before run-time the execution sequence of a program, and bec(o.use these techniques introduce very little overhead at run-time (due to simplicity and hardware implementation). Additionally, the techniques used for scheduling and data distribution are distributed to avoid serialization.","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"130 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122795835","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633082
M. Gupta, P. Banerjee
Abstract : An important problem facing numerous research projects on parallelizing compilers for distributed memory machines is that of automatically determining a suitable data partitioning scheme for a program. Most of the current projects leave this tedious problem almost entirely to the user. In this paper, we present a novel approach to the problem of automatic data partitioning. We introduce the notion of constraints on data distribution, and show how a parallelizing compiler can infer those constraints by looking at the data reference patterns in the source code of the program. We show how these constraints may be combined by the compiler to obtain a complete and consistent picture of the data distribution scheme, one that offers good performance in terms of the overall execution time. We illustrate our approach on an example routine, TRED2, from the EISPACK library, to demonstrate its applicability to real programs. Finally, we discuss briefly some other approaches that have recently been proposed for this problem, and argue why ours seems to be more general and powerful.
{"title":"Automatic Data Partitioning on Distributed Memory Multiprocessors","authors":"M. Gupta, P. Banerjee","doi":"10.1109/DMCC.1991.633082","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633082","url":null,"abstract":"Abstract : An important problem facing numerous research projects on parallelizing compilers for distributed memory machines is that of automatically determining a suitable data partitioning scheme for a program. Most of the current projects leave this tedious problem almost entirely to the user. In this paper, we present a novel approach to the problem of automatic data partitioning. We introduce the notion of constraints on data distribution, and show how a parallelizing compiler can infer those constraints by looking at the data reference patterns in the source code of the program. We show how these constraints may be combined by the compiler to obtain a complete and consistent picture of the data distribution scheme, one that offers good performance in terms of the overall execution time. We illustrate our approach on an example routine, TRED2, from the EISPACK library, to demonstrate its applicability to real programs. Finally, we discuss briefly some other approaches that have recently been proposed for this problem, and argue why ours seems to be more general and powerful.","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"357 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125644906","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633104
T. Znati, R. Melhem, K. Pruhs
This paper is concerned with sender-initiated load balancing algorithms for parallel architecture that take into consideration the dilation between the sender and the receiver of the migrated task. The basic scheme proposed in this paper, uses a load contention number that accounts for the dilation among processors. This mechanism is generalized to reflect the specijic requirements of different environments. W e also describe variations of the basic scheme that aim at reducing the interaction overhead among contending processors.
{"title":"Dilation Based Bidding Schemes For Dynamic Load Balancing On Distributed Processing system","authors":"T. Znati, R. Melhem, K. Pruhs","doi":"10.1109/DMCC.1991.633104","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633104","url":null,"abstract":"This paper is concerned with sender-initiated load balancing algorithms for parallel architecture that take into consideration the dilation between the sender and the receiver of the migrated task. The basic scheme proposed in this paper, uses a load contention number that accounts for the dilation among processors. This mechanism is generalized to reflect the specijic requirements of different environments. W e also describe variations of the basic scheme that aim at reducing the interaction overhead among contending processors.","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121642109","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633305
Jianping Zhu
A parallel numericail algorithm and its inaplementation on intel hypercubes are presented in this paper. Numerical computations show that the algorithm presented here can be used to handle reservoir models with hundreds of thousand gr id points. The block SOR algorithm with red-black ordering is used to solve the matrix equations in parallel and the binary reflection gray code algorithm is used to embed a ring structure into a hypercube to secure that the internode communication always occurs between the nearest neighbors. The efficiency of the algorithm can be well over 90 percent for the right problem site and number of nodes.
{"title":"Oil Reservoir Simulations on Hypercubes","authors":"Jianping Zhu","doi":"10.1109/DMCC.1991.633305","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633305","url":null,"abstract":"A parallel numericail algorithm and its inaplementation on intel hypercubes are presented in this paper. Numerical computations show that the algorithm presented here can be used to handle reservoir models with hundreds of thousand gr id points. The block SOR algorithm with red-black ordering is used to solve the matrix equations in parallel and the binary reflection gray code algorithm is used to embed a ring structure into a hypercube to secure that the internode communication always occurs between the nearest neighbors. The efficiency of the algorithm can be well over 90 percent for the right problem site and number of nodes.","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127895051","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633091
A. Boals, A. Gupta, J. Hashmi, N. Sherwani
{"title":"Efficient Algorithms for Scheduling Tasks on Hypercubes","authors":"A. Boals, A. Gupta, J. Hashmi, N. Sherwani","doi":"10.1109/DMCC.1991.633091","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633091","url":null,"abstract":"","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116649913","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633359
R. Rowley, B. Bose
We show that a slightly modified degree 2r de Bruijn graph can be decomposed into r Hamiltonian cycles when r is a power of a prime. Adjacent nodes in the de Bruijn graph remain adjacent in the modified graph, and the maximum degree does not increase. The presence of edge-disjoint Hamiltonian cycles provides an advantage when implementing algorithms that requ.ire a ring structure by allowing message traflc to be spread evenly across the network. The changes also enhance fault tolerance.
{"title":"Edge-Disjoint Hamiltonian Cycles in de Bruijn Networks","authors":"R. Rowley, B. Bose","doi":"10.1109/DMCC.1991.633359","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633359","url":null,"abstract":"We show that a slightly modified degree 2r de Bruijn graph can be decomposed into r Hamiltonian cycles when r is a power of a prime. Adjacent nodes in the de Bruijn graph remain adjacent in the modified graph, and the maximum degree does not increase. The presence of edge-disjoint Hamiltonian cycles provides an advantage when implementing algorithms that requ.ire a ring structure by allowing message traflc to be spread evenly across the network. The changes also enhance fault tolerance.","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124024241","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 : 1991-04-28DOI: 10.1109/DMCC.1991.633208
T. Huntsberger
Since the form of the nuclear force is not explicitly known, various models have been proposed to account for the nucleon-nucleon scattering data and gross properties of nuclei, such as quadrupole moments, binding energy, etc. Some of the better known models are the liquid-drop [7], shell [4], unified [5] and grouptheoretical [6] approximations. This paper presents a hypercube based algorithm for the derivation of nuclear structure energy levels and binding energies for a wide range o isotopes. The algorithm is based on previous work f l, 21 which studied isotopes in the f-p-g shell shown in Figure 1. Extensions to higher order shells are obtained using a distributed decomposition of the coupled state and perturbation calculations. The results of scaling experiments for the NCUBE/lO system at the University of South Carolina are also included.
{"title":"Distributed algorithm for nuclear structure investigations","authors":"T. Huntsberger","doi":"10.1109/DMCC.1991.633208","DOIUrl":"https://doi.org/10.1109/DMCC.1991.633208","url":null,"abstract":"Since the form of the nuclear force is not explicitly known, various models have been proposed to account for the nucleon-nucleon scattering data and gross properties of nuclei, such as quadrupole moments, binding energy, etc. Some of the better known models are the liquid-drop [7], shell [4], unified [5] and grouptheoretical [6] approximations. This paper presents a hypercube based algorithm for the derivation of nuclear structure energy levels and binding energies for a wide range o isotopes. The algorithm is based on previous work f l, 21 which studied isotopes in the f-p-g shell shown in Figure 1. Extensions to higher order shells are obtained using a distributed decomposition of the coupled state and perturbation calculations. The results of scaling experiments for the NCUBE/lO system at the University of South Carolina are also included.","PeriodicalId":313314,"journal":{"name":"The Sixth Distributed Memory Computing Conference, 1991. Proceedings","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129377627","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：