Zhihui Du, Sanli Li, Shuyou Li, Mengyue Wu, Jing Zhu
{"title":"Massively parallel simulated annealing embedded with downhill-a SPMD algorithm for cluster computing","authors":"Zhihui Du, Sanli Li, Shuyou Li, Mengyue Wu, Jing Zhu","doi":"10.1109/IWCC.1999.810899","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810899","url":null,"abstract":"Simulated Annealing (SA) is a frequently used stochastic algorithm to deal with combinatorial optimization problems and it converges with probability infinitely close to 1. SA is an NP algorithm and the long executive time prevents it from being accepted for many real-time applications. This paper presents a SPMD (Single Program Multiple Data) algorithm which combines SA with local searching algorithm-downhill. The hybrid method not only keeps the convergence of SA but also improves the convergence speed of SA. Approximate solutions can be found quickly for complex optimization problems and more precise solutions can also be found by employing the same algorithm to fine-tune the approximate solutions. SA is an essential serial algorithm, but the SPMD algorithm breaks up the serial bottleneck of SA and its performance scales up linearly with the increase of processors, at the same time, the SPMD algorithm does not require careful choice of control parameters. Application cases show that the algorithm is robust and it can find high quality solution with high speed.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"212 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121721319","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Global resource director (GRD) customer scenarios for large multiprocessor environments","authors":"F. Ferstl","doi":"10.1109/IWCC.1999.810904","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810904","url":null,"abstract":"Genias Global Resource Director (GRD) was developed to meet the demands of large multiprocessor sites with several 100s of CPUs and users. At the Army Research Laboratory (ARL) and at BMW, GRD is installed on several types of Cray and SGI machines. This paper will describe how GRD integrates such resources into one environment automatically managed by global resource utilization policies. Benefits that the sites achieved in using GRD such as avoiding over-subscription of resources, automated management and utilization monitoring will also be discussed. In addition, an overview will be provided on the groundbreaking functionality of GRD which features a ticketing system based distribution of resource shares combined with online control of the CPU utilization of running jobs (also known as \"dynamic scheduling\").","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122819647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Realistic communication model for parallel computing on cluster","authors":"Anthony T. C. Tam, Cho-Li Wang","doi":"10.1109/IWCC.1999.810813","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810813","url":null,"abstract":"We present a model for parallel computation on commodity cluster. Our cluster model is targeted as a tool for performance analysis and algorithm design. We abstract the communication event by means of local and remote data movements, and explicitly expose the contention problems by capturing them in our parameters. To validate our model, we compare the prediction accuracy of our model with the Postal model for the popular tree-based broadcast algorithm. Our model provides good prediction accuracy and answers to some performance issues that are missing in existing models. We examine the gather collective operation, in which contention delay dominates its overall execution time. Based on the model, we design a communication schedule for the gather operation that is based on the upper and lower bounds, in which the congestion behavior could be under our control and achieve optimal results by avoiding data loss.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"94 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131369632","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Building and scheduling parallel adaptive applications in heterogeneous environments","authors":"D. Kebbal, E. Talbi, J. Geib","doi":"10.1109/IWCC.1999.810825","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810825","url":null,"abstract":"In this paper we present a dynamic approach for constructing and scheduling parallel adaptive applications in heterogeneous multi-user environments (networks of workstations). Parallel adaptive applications have the property of varying their parallelism degree following the load fluctuation of the underlying environment. Our tool provides a programming facility that allows the application construction to avoid managing these complex problems and an allocation module responsible for running and scheduling application tasks. The allocation module handles also all problems related to the dynamic character of the application so that the user may not know at any time whether his application executes on one or dozens of workstations. The allocation module is completed by a scheduler which tries to make good mapping decisions and to adjust the mapping when the application reconfigures dynamically. The scheduling approach based on the dependency graphs model tries to minimize the execution time of the application by decreasing the parallelism loss situations in which some nodes allocated to the application are waiting for the work availability which must be generated by some slow nodes. This can be achieved by analysing dynamically the dep-graph structure and using the heterogeneity aspect. Encouraging results were obtained from experiments conducted on a parallel version of the Gaussian elimination application which is not well adapted to our environment.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"817 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133004760","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A finite element solver for convection diffusion problems using a cluster of PCs","authors":"R. Silva","doi":"10.1109/IWCC.1999.810902","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810902","url":null,"abstract":"In this work an experimental cluster of PCs is presented. Some benchmarks and a Krylov-Schwarchz solver for convection diffusion problems are considered. The results obtained are stimulating and point out the necessity of improving the topology configuration.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128118221","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Sharing the Garden GATE: towards an efficient uniform programming model for CLUMPS","authors":"D. Butler, P. Roe","doi":"10.1109/IWCC.1999.810896","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810896","url":null,"abstract":"Gardens integrates a programming language and system to support parallel computation over a network of workstations. Our goal is to develop extensions to Gardens allowing it to efficiently support clusters of SMP machines under a uniform programming model. Such support requires the implementation of high-performance shared memory message passing primitives as well as changes to the existing model invariants and the Gardens runtime system itself: The new communication primitives have demonstrated a minimum latency of 2.0 microseconds and can maintain a maximum bandwidth of approximately 194 MB/s on our test system. The NAS benchmarks, EP and CG have also demonstrated significant performance increases when parallelised by Gardens. The major contribution of this work is a system that efficiently supports multiprocessor computers in a network of workstations with a uniform programming model. Such a system promises to greatly simplify the task of developing high-performance parallel applications to run on CLUMPS.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131352735","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"HP Hyperplex(R) clustering technology","authors":"Daniel P. S. Charlu","doi":"10.1109/IWCC.1999.810905","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810905","url":null,"abstract":"HP HyperPlex(R) is a clustering technology that leverages HP platforms, mass storage and interconnect products to provide customers with mission-critical high availability and increased processing capacity. A HyperPlex cluster consists of two or more symmetric multiprocessing (SMP) nodes, an interconnect, mass storage options, and a centrally-managed operating environment. Further, HyperPlex solutions lower costs by consolidating resources and increasing users' productivity. This paper describes the technologies for each of these components, the features and benefits of Hewlett-Packard's HyperPlex implementation, and the software that complements these technologies.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114953139","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}
T. Monteil, Jean-Marie Garcia, David Gauchard, O. Brun
{"title":"Communication kernel for high speed networks in the parallel environment LANDA-HSN","authors":"T. Monteil, Jean-Marie Garcia, David Gauchard, O. Brun","doi":"10.1109/IWCC.1999.810810","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810810","url":null,"abstract":"Due to the development of several new technologies concerning high speed networks, one can get PC clusters for intensive computing at a low price. Parallel programming environments must take into account the heterogeneity of workstation networks into their communication kernels. New technologies must be easily integrated into such environments. The LANDA-HSN project has been designed for an efficient use of the available resources (network, processor, memory). The paper describes the mechanisms and the implementation of its communication kernel on SMP hosts and through interconnection networks. The contribution of the paper is to propose a flexible and efficient communication kernel.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"99 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125777953","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Single I/O space for scalable cluster computing","authors":"Roy S. C. Ho, Hai Jin, K. Hwang","doi":"10.1109/IWCC.1999.810821","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810821","url":null,"abstract":"In this paper, we propose a novel single I/O space architecture for achieving a single system image (SSI) at the I/O subsystem level. This is very much desired in a scalable cluster computing environment using commodity components. Our design achieves a single address space for all blocks of data in the cluster, which can tolerate all single disk failures. While traditional approaches focused on at user-level or at the distributed file subsystem level, we separate the I/O subsystem of a cluster into the file system and a set of distributed virtual device drivers (VDDs). All SSI services are provided by the VDDs with unmodified file systems. Compared to previous approaches, our approach has higher transparency, better performance, lower implementation cost, higher availability, and application compatibility for I/O intensive cluster computing.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124190378","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The influence of concurrent process duplication on the performance of parallel applications executing on COWs","authors":"M. Hobbs, A. Goscinski","doi":"10.1109/IWCC.1999.810898","DOIUrl":"https://doi.org/10.1109/IWCC.1999.810898","url":null,"abstract":"The benefits gained from utilising Clusters of Workstations (COWs) as architectures for parallel processing are well known with such systems being a common sight throughout many organisations. COWs are able to present users and application programmers with a considerable computational resource for a relatively low cost. Unfortunately, the software currently used to support the application programmer in the development and execution of their parallel programs is often limited and lacks transparency. The initialisation stage of a parallel program can prove critical in obtaining good performance and engaging programmers, especially as the number of parallel processes increases. The focus of this paper is to present a unique process instantiation service that allows a set of parallel child processes to be concurrently duplicated across a COW through the use of group process migration and group communication services. The importance and impact that this process duplication service has on some common parallel applications is presented and detailed.","PeriodicalId":276367,"journal":{"name":"ICWC 99. IEEE Computer Society International Workshop on Cluster Computing","volume":"16 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1999-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123266195","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}