{"title":"An empirical approach for efficient all-to-all personalized communication on Ethernet switched clusters","authors":"Ahmad Faraj, Xin Yuan","doi":"10.1109/ICPP.2005.19","DOIUrl":"https://doi.org/10.1109/ICPP.2005.19","url":null,"abstract":"All-to-all personalized communication (AAPC) is one of the most commonly used communication patterns in parallel applications. Developing an efficient AAPC routine is difficult since many system parameters can affect the performance of an AAPC algorithm. In this paper, we investigate an empirical approach for automatically generating efficient AAPC routines for Ethernet switched clusters. This approach applies when the application execution environment is decided, and it allows efficient customized AAPC routines to be created. Experimental results show that the empirical approach generates routines that consistently achieve high performance on clusters with different network topologies. In many cases, the automatically generated routines out-perform conventional AAPC implementations to a large degree.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116935745","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":"PFED: a prediction-based fair active queue management algorithm","authors":"Wenyu Gao, Jianxin Wang, Jianer Chen, Songqiao Chen","doi":"10.1109/ICPP.2005.63","DOIUrl":"https://doi.org/10.1109/ICPP.2005.63","url":null,"abstract":"In this paper, we propose a novel active queue management algorithm PFED, which is based on network traffic prediction. The main properties of PFED are: (1) stabilizing queue length at a desirable level with consideration of future traffic, and using a MMSE (minimum mean square error) predictor to predict future network traffic; (2) imposing effective punishment upon misbehaving flow with a full stateless method; (3) maintaining queue arrival rate bounded by queue service rate through more reasonable calculation of packet drop probability. To verify the performance of PFED, PFED is implemented in NS2 and is compared with RED and CHOKe with respect to different performance metrics. Simulation results show that PFED outperforms RED and CHOKe in stabling instantaneous queue length and in fairness. It is also shown that PFED enables the link capacity to be fully utilized by stabilizing the queue length at a desirable level, while not incurring excessive packet loss ratio.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123657053","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":"Service migration in distributed virtual machines for adaptive grid computing","authors":"S. Fu, Chengzhong Xu","doi":"10.1109/ICPP.2005.71","DOIUrl":"https://doi.org/10.1109/ICPP.2005.71","url":null,"abstract":"Computational grids can integrate geographically distributed resources into a seamless environment. To facilitate managing these heterogeneous resources, the virtual machine technology provides a powerful layer of abstraction and allows multiple applications to multiplex the resources of a grid computer. On the other hand, the grid dynamics requires the virtual machine system be distributed and reconfigurable. However, the existing migration approaches only move the execution entities, such as processes, threads, and mobile agents, among servers and leave the runtime services behind. They are not potent to achieve service reconfiguration in face of server overload or failures. In this paper, we propose a service migration mechanism, which moves the computational services of a virtual server, for instance a shared array runtime support system, to available servers for adaptive grid computing. In this way, parallel jobs can resume computation on a remote server without requiring service preinstallation. As an illustration of the service migration mechanism, we incorporated it into a Java-compliant distributed virtual machine, DSA, and formed a Mobile DSA (M-DSA) to accommodate adaptive parallel applications in grids. We measured the performance of M-DSA in the execution of applications from the SPLASH-2 benchmark suite on a campus grid. Experimental results show that service migration can achieve system adaptivity effectively.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"129 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120863236","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":"Design tradeoffs for BLAS operations on reconfigurable hardware","authors":"Ling Zhuo, V. Prasanna","doi":"10.1109/ICPP.2005.31","DOIUrl":"https://doi.org/10.1109/ICPP.2005.31","url":null,"abstract":"Numerical linear algebra operations are key primitives in scientific computing. Performance optimizations of such operations have been extensively investigated and some basic operations have been implemented as software libraries. With the rapid advances in technology, hardware acceleration of linear algebra applications using FPGAs (field programmable gate arrays) has become feasible. In this paper, we propose FPGA-based designs for several BLAS operations, including vector product, matrix-vector multiply, and matrix multiply. By identifying the design parameters for each BLAS operation, we analyze the design tradeoffs. In the implementations of the designs, the values of the design parameters are determined according to the hardware constraints, such as the available area, the size of on-chip memory, the external memory bandwidth and the number of I/O pins. The proposed designs are implemented on a Xilinx Virtex-II Pro FPGA.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"76 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114066696","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 peer-to-peer replica management service for high-throughput grids","authors":"A. Chazapis, A. Zissimos, N. Koziris","doi":"10.1109/ICPP.2005.12","DOIUrl":"https://doi.org/10.1109/ICPP.2005.12","url":null,"abstract":"Future high-throughput grids may integrate millions or even billions of processing and data storage nodes. Services provided by the underlying grid infrastructure may have to be able to scale to capacities not even imaginable today. In this paper we concentrate on one of the core components of the data grid architecture - the replica location service - and evaluate a redesign of the system based on a structured peer-to-peer network overlay. We argue that the architecture of the currently most widespread solution for file replica location on the grid is biased towards high-performance deployments and cannot scale to the future needs of a global grid. Structured peer-to-peer systems can provide the same functionality, while being much more manageable, scalable and fault-tolerant. However, they are only capable of storing read-only data. To this end, we propose a revised protocol for distributed hash tables that allows data to be changed in a distributed and scalable fashion. Results from a prototype implementation of the system suggest that grids can truly benefit from the scalability and fault-tolerance properties of such peer-to-peer algorithms.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"190 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121624308","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 smart TCP socket for distributed computing","authors":"Shao Tao, A. Ananda","doi":"10.1109/ICPP.2005.14","DOIUrl":"https://doi.org/10.1109/ICPP.2005.14","url":null,"abstract":"The middleware in distributed computing coordinates a group of servers to accomplish a resource intensive task; however, without resource monitoring, the server selection algorithms will not be sophisticated enough to provide satisfying results at all time. This paper presents a smart TCP socket library using server workload feedback to improve selection techniques. Users are able to specify server requirements by using a predefined meta language. Monitoring components such as the server probes and monitors are in charge of monitoring the server resource usage, network metrics and performing security measurements. Based on the user's request and available server resources, the user request handler called wizard will recommend the most suitable servers to user applications. The centralized and the distributed modes are both provided so that the socket library can be adapted to both small distributed systems and a large scale grid.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133747317","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":"Toward effective NIC caching: a hierarchical data cache architecture for iSCSI storage servers","authors":"Xiaoyu Yao, Jun Wang","doi":"10.1109/ICPP.2005.76","DOIUrl":"https://doi.org/10.1109/ICPP.2005.76","url":null,"abstract":"In this paper, we present a hierarchical data cache architecture called DCA to effectively slash local interconnect traffic and thus boost the storage server performance. DCA is composed of a read cache in NIC card called NIC cache and a read/write unify cache in host memory called Helper cache. NIC cache services most portion of read requests without fetching data via PCI bus, while Helper cache 1) supplies some portions of read requests given partial NIC cache hits; 2) directs cache placement for NIC cache and 3) absorbs most transient writes locally. We developed a novel State Locality Aware cache Placement algorithm called SLAP to improve NIC cache hit ratio for mixed read and write workloads. To demonstrate the effectiveness of DCA, we developed a DCA prototype system and evaluated it with open source iSCSI implementation under representative storage server workloads. Experimental results showed that DCA can boost iSCSI storage server throughput by up to 121% and slash the PCI traffic by up to 74% compared with an iSCSI target without DCA.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"18 4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130905975","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":"Fast barriers for scalable ccNUMA systems","authors":"Liqun Cheng, J. Carter","doi":"10.1109/ICPP.2005.39","DOIUrl":"https://doi.org/10.1109/ICPP.2005.39","url":null,"abstract":"The contributions of this paper are threefold. First, we identify and quantify the performance deficiencies of conventional barrier implementations when they are executed on real (non-idealized) hardware. Second, we propose a queue-based barrier algorithm that has effectively O(1) time complexity as measured in round trip message latencies. Third, we demonstrate how matching the barrier implementation to the way that modern shared memory systems operate can improve performance dramatically by exploiting a hardware write-update (PUT) mechanism for signaling. The resulting barrier algorithm only costs one serialized round trip message latency to perform a barrier operation across N processors. Using a cycle-accurate execution-driven simulator of a future-generation SGI multiprocessor, we show that with no special hardware support our queue-based barrier outperforms OpenMP's LL/SC-based barrier implementation by a factor of 7.9 on 256 processors. With hardware that supports a coherent PUT operation, our queue-based barrier outperforms OpenMP barriers by a factor of 94 and outperforms barriers based on SGI's memory controller-based atomic operations by a factor of 6.5 on 256 processors.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131886115","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":"Considering the relative importance of network performance and network features","authors":"W. Lawry, K. Underwood","doi":"10.1109/ICPP.2005.26","DOIUrl":"https://doi.org/10.1109/ICPP.2005.26","url":null,"abstract":"Latency and bandwidth are usually considered to be the dominant factor in parallel application performance; however, recent studies have indicated that support for independent progress in MPI can also have a significant impact on application performance. This paper leverages the Cplant system at Sandia National Labs to compare a faster, vendor provided MPI library without independent progress to an internally developed MPI library that sacrifices some performance to provide independent progress. The results are surprising. Although some applications see significant negative impacts from the reduced network performance, others are more sensitive to the presence of independent progress.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"164 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114435813","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":"Stimulus-based adaptive sleeping for wireless sensor networks","authors":"Hoilun Ngan, Yanmin Zhu, L. Ni, Renyi Xiao","doi":"10.1109/ICPP.2005.74","DOIUrl":"https://doi.org/10.1109/ICPP.2005.74","url":null,"abstract":"Wireless sensor networks have been widely deployed for monitoring environments of interest. The diffusion stimulus (DS) is a very common and important one in real environments, which is characterized by originating at a source spot and continuously spreading outward. DS-based applications are usually time-sensitive and require high accuracy. It is very challenging, however, for sensor networks to monitor a DS when those sensors could usually monitor one point (the sensing range is zero) only. No existing algorithm is effective and energy-efficient for DS monitoring. In this paper, we propose stimulus-based adaptive sleeping (SAS), to tackle this unique challenge. With SAS, each node independently determines its sleep duration based on its local observations on the stimulus. SAS enables sensors near the stimulus boundary to stay alert to accurately capture the stimulus arrival, and those far away from the stimulus to safely sleep longer for energy efficiency.","PeriodicalId":440948,"journal":{"name":"2005 International Conference on Parallel Processing (ICPP'05)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116054781","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}