Pub Date : 1994-05-02DOI: 10.1109/MPCS.1994.367062
N. Rishe, A. Shaposhnikov, Wei Sun
We are developing a massively parallel semantic database machine. Our basic semantic storage structure insures balanced load for most parts of the database. For the other parts of the database, a load balancing algorithm is proposed herein, which allows inexpensive dynamic rebalancing without substantial negative impact on queries and transactions.<>
{"title":"Load balancing policy in a massively parallel semantic database","authors":"N. Rishe, A. Shaposhnikov, Wei Sun","doi":"10.1109/MPCS.1994.367062","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367062","url":null,"abstract":"We are developing a massively parallel semantic database machine. Our basic semantic storage structure insures balanced load for most parts of the database. For the other parts of the database, a load balancing algorithm is proposed herein, which allows inexpensive dynamic rebalancing without substantial negative impact on queries and transactions.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"10 1","pages":"328-332"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75746601","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367075
Z. S. Khan, E. Kwatny
On a general purpose massively parallel system such as the Connection Machine (CM-2), the performance of the hashing algorithm is largely dependent on the interprocessor communication architecture. This paper compares the performance of hashing data using the router and NEWS communication mechanisms on the CM-2, Although in general, the NEWS communication mechanism is faster than router communication, our results show that for hashing data from uniform, normal, and normal distribution with a high peak, the router communication outperforms NEWS communication.<>
{"title":"Performance comparison of interprocessor communication schemes for hashing techniques on the Connection Machine","authors":"Z. S. Khan, E. Kwatny","doi":"10.1109/MPCS.1994.367075","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367075","url":null,"abstract":"On a general purpose massively parallel system such as the Connection Machine (CM-2), the performance of the hashing algorithm is largely dependent on the interprocessor communication architecture. This paper compares the performance of hashing data using the router and NEWS communication mechanisms on the CM-2, Although in general, the NEWS communication mechanism is faster than router communication, our results show that for hashing data from uniform, normal, and normal distribution with a high peak, the router communication outperforms NEWS communication.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"9 1","pages":"211-215"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76590826","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367087
J. Padget
A multicomputer is a number of interconnected heterogeneous computing systems being used as a single computing system. To factor out the actual number and kinds of the resources available in any one (physical) multicomputer we propose the virtual multicomputer in which an abstract archetype of each kind (workstation, shared-memory multiprocessor, distributed-memory multiprocessor, array processor) can be instantiated at will. In turn, this raises the issues of how to control such a machine and, a fortiori, how to communicate within such a machine. While these problems can readily be solved in each program by using low-level techniques, it seems desirable to search for higher level abstractions which capture patterns of communication and control in concurrent programs in an architecture-independent manner. We outline some steps in this direction based on the use of EuLisp and its object system in a distributed environment.<>
{"title":"Controlling (virtual) multicomputers","authors":"J. Padget","doi":"10.1109/MPCS.1994.367087","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367087","url":null,"abstract":"A multicomputer is a number of interconnected heterogeneous computing systems being used as a single computing system. To factor out the actual number and kinds of the resources available in any one (physical) multicomputer we propose the virtual multicomputer in which an abstract archetype of each kind (workstation, shared-memory multiprocessor, distributed-memory multiprocessor, array processor) can be instantiated at will. In turn, this raises the issues of how to control such a machine and, a fortiori, how to communicate within such a machine. While these problems can readily be solved in each program by using low-level techniques, it seems desirable to search for higher level abstractions which capture patterns of communication and control in concurrent programs in an architecture-independent manner. We outline some steps in this direction based on the use of EuLisp and its object system in a distributed environment.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"14 1","pages":"102-112"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79511241","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367084
J. Gaber, B. Toursel, G. Goncalves, T. Hsu
Trees are an important class of computational structures due to their favourable property of logarithmic path from the root to any leaf element. Mapping problems arise when the tree structure differs from the processor interconnection of the parallel computer. The MasPar's MP-2 is a massively parallel computer where processors are interconnected via the X-Net neighborhood two-dimensional mesh and the global multistage crossbar router network. Novel mapping schemes for trees on the MasPar's MP-2 two-dimensional mesh and on the two-dimensional mesh together with the MasPar's MP-2 multistage crossbar network are proposed here. Appropriate algorithms of the mapping scheme on the two-dimensional mesh (or grid) are presented and shown to be superior over known mappings on square array (or grids). These mapping schemes are well suited for massively parallel systems which focus on data parallelism. This parallelism comes from simultaneous operations across large sets of data. It is the data parallelism style which opposes the control parallel style used in multiprocessing.<>
{"title":"Embedding tree structures in MPCs: application to the MasPar MP-2","authors":"J. Gaber, B. Toursel, G. Goncalves, T. Hsu","doi":"10.1109/MPCS.1994.367084","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367084","url":null,"abstract":"Trees are an important class of computational structures due to their favourable property of logarithmic path from the root to any leaf element. Mapping problems arise when the tree structure differs from the processor interconnection of the parallel computer. The MasPar's MP-2 is a massively parallel computer where processors are interconnected via the X-Net neighborhood two-dimensional mesh and the global multistage crossbar router network. Novel mapping schemes for trees on the MasPar's MP-2 two-dimensional mesh and on the two-dimensional mesh together with the MasPar's MP-2 multistage crossbar network are proposed here. Appropriate algorithms of the mapping scheme on the two-dimensional mesh (or grid) are presented and shown to be superior over known mappings on square array (or grids). These mapping schemes are well suited for massively parallel systems which focus on data parallelism. This parallelism comes from simultaneous operations across large sets of data. It is the data parallelism style which opposes the control parallel style used in multiprocessing.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"47 1","pages":"132-139"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77804744","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367067
E. Pissaloux, P. Bonnin
Finding the best adequacy of architectures and algorithms for low and intermediate image processing is the central problem of design of any system dedicated to image processing. This paper analyses some image processing algorithms in order to define pertinent invariants in image processing. It includes the data types used, and the evolution of their representation during the processing and data movements. The common, but essential, characteristics are used for the definition of the most suitable architectural invariants-basic elements for any machine dedicated to image processing. Some implementations of several invariants on some existing massively parallel architectures are also discussed.<>
{"title":"On the adequacy of image processing algorithms and massively parallel computers","authors":"E. Pissaloux, P. Bonnin","doi":"10.1109/MPCS.1994.367067","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367067","url":null,"abstract":"Finding the best adequacy of architectures and algorithms for low and intermediate image processing is the central problem of design of any system dedicated to image processing. This paper analyses some image processing algorithms in order to define pertinent invariants in image processing. It includes the data types used, and the evolution of their representation during the processing and data movements. The common, but essential, characteristics are used for the definition of the most suitable architectural invariants-basic elements for any machine dedicated to image processing. Some implementations of several invariants on some existing massively parallel architectures are also discussed.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"39 1","pages":"282-294"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83864282","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367080
C. Alippi, V. Piuri, M. Sami
The problem of sensitivity to errors in artificial neural networks is discussed here in behavioral terms, i.e. considering an abstract model of the network and the errors that can affect a neuron's computation. Feed-forward multi-layered networks are considered; the performance taken into account with respect to error sensitivity is their classification capacity. The final aim is evaluation of the probability that a single neuron's error will affect both its own classification capacity and the whole network's classification capacity. A geometrical representation of the neural computation is adopted as the basis for such evaluation. Probability of error propagation is evaluated with respect to the single neuron's output as well as to the complete network's output. The information derived as used to evaluate, for a specific digital network architecture, the must critical sections of the implementation as far as reliability is concerned and thus to point out candidates for ad-hoc fault-tolerance policies.<>
{"title":"The issue of error sensitivity in neural networks","authors":"C. Alippi, V. Piuri, M. Sami","doi":"10.1109/MPCS.1994.367080","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367080","url":null,"abstract":"The problem of sensitivity to errors in artificial neural networks is discussed here in behavioral terms, i.e. considering an abstract model of the network and the errors that can affect a neuron's computation. Feed-forward multi-layered networks are considered; the performance taken into account with respect to error sensitivity is their classification capacity. The final aim is evaluation of the probability that a single neuron's error will affect both its own classification capacity and the whole network's classification capacity. A geometrical representation of the neural computation is adopted as the basis for such evaluation. Probability of error propagation is evaluated with respect to the single neuron's output as well as to the complete network's output. The information derived as used to evaluate, for a specific digital network architecture, the must critical sections of the implementation as far as reliability is concerned and thus to point out candidates for ad-hoc fault-tolerance policies.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"113 1","pages":"177-189"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90661228","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367051
Hans-Ulrich Heiß
We consider multiprogramming operation in large scale grid-connected multicomputer systems where the grid is spatially partitioned among the programs. We regard the problem as a two-dimensional resource allocation problem for which different algorithms are proposed. Results from simulation experiments indicate the performance that can be achieved in terms of fragmentation and throughput.<>
{"title":"Dynamic partitioning of large multicomputer systems","authors":"Hans-Ulrich Heiß","doi":"10.1109/MPCS.1994.367051","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367051","url":null,"abstract":"We consider multiprogramming operation in large scale grid-connected multicomputer systems where the grid is spatially partitioned among the programs. We regard the problem as a two-dimensional resource allocation problem for which different algorithms are proposed. Results from simulation experiments indicate the performance that can be achieved in terms of fragmentation and throughput.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"50 1","pages":"413-417"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78413883","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367050
J. Kwiatkowski, M. Andruszkiewicz, S. Plaskowicki
Presently there is the need for incorporating novel issues, for example new hardware architectures like array processors, dataflow computers, and new concepts like nondeterminism and concurrency into existing computer science curricula. Dataflow processing is based on the principle that an instruction is executable as soon as all its operands value are available. Dataflow will thus take maximum advantage of the parallelism in the program without having to specify this parallelism explicitly. The paper presents a short description of the integrated software environment which gives the possibility of programming dataflow computers at different programming levels.<>
{"title":"Integrated software environment for dataflow computation","authors":"J. Kwiatkowski, M. Andruszkiewicz, S. Plaskowicki","doi":"10.1109/MPCS.1994.367050","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367050","url":null,"abstract":"Presently there is the need for incorporating novel issues, for example new hardware architectures like array processors, dataflow computers, and new concepts like nondeterminism and concurrency into existing computer science curricula. Dataflow processing is based on the principle that an instruction is executable as soon as all its operands value are available. Dataflow will thus take maximum advantage of the parallelism in the program without having to specify this parallelism explicitly. The paper presents a short description of the integrated software environment which gives the possibility of programming dataflow computers at different programming levels.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"56 1","pages":"418-422"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77862343","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367033
M. Furnari, A. Massarotti
In this paper we review symbolic parallel programming issues. We start describing the peculiarities of symbolic programming. Next, we review the standard parallel programming framework for the imperative-like languages. Also the autoscheduling environment, proposed by Polychronopoulos (1990), is introduced. The current extensions for symbolic parallel programming are described. The problem how to efficiently manage the parallelism in LISP is addressed, and the Hierarchical Task Graph, an intermediate program representation in which the parallelism is expressed by the task hierarchy, and the associated run-time model are also described. The extensions necessary to manage the parallelism at run-time in a multiprogramming environment are described.<>
{"title":"Symbolic computing, Lisp languages, and parallel computing","authors":"M. Furnari, A. Massarotti","doi":"10.1109/MPCS.1994.367033","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367033","url":null,"abstract":"In this paper we review symbolic parallel programming issues. We start describing the peculiarities of symbolic programming. Next, we review the standard parallel programming framework for the imperative-like languages. Also the autoscheduling environment, proposed by Polychronopoulos (1990), is introduced. The current extensions for symbolic parallel programming are described. The problem how to efficiently manage the parallelism in LISP is addressed, and the Hierarchical Task Graph, an intermediate program representation in which the parallelism is expressed by the task hierarchy, and the associated run-time model are also described. The extensions necessary to manage the parallelism at run-time in a multiprogramming environment are described.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"43 1","pages":"542-553"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89905808","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 : 1994-05-02DOI: 10.1109/MPCS.1994.367078
A. Bellaachia, S. Berkovich, A. Meltzer, Z. Kitov
This paper presents a new pipelined microprocessor-based architecture (PMA) using On-The Fly processing. The effectiveness of the PMA has been investigated for several applications such as database operations and very large databases using WORM (Write-Once-Read-Many) optical disk technology. In this paper, a pipelined information retrieval algorithm and a pipelined join algorithm are presented and their complexities are analyzed. Extensive simulations of these algorithms have been conducted on an emulated PMA on an AT&T parallel machine. Parallel implementations of these algorithms have been also considered for comparison purposes.<>
{"title":"Pipelined microprocessor-based architecture: on-the-fly processing","authors":"A. Bellaachia, S. Berkovich, A. Meltzer, Z. Kitov","doi":"10.1109/MPCS.1994.367078","DOIUrl":"https://doi.org/10.1109/MPCS.1994.367078","url":null,"abstract":"This paper presents a new pipelined microprocessor-based architecture (PMA) using On-The Fly processing. The effectiveness of the PMA has been investigated for several applications such as database operations and very large databases using WORM (Write-Once-Read-Many) optical disk technology. In this paper, a pipelined information retrieval algorithm and a pipelined join algorithm are presented and their complexities are analyzed. Extensive simulations of these algorithms have been conducted on an emulated PMA on an AT&T parallel machine. Parallel implementations of these algorithms have been also considered for comparison purposes.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"55 1","pages":"196-200"},"PeriodicalIF":0.0,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86265407","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：