the proceedings contain 40 papers. the topics discussed include: a classified cluster based architecture for resource discovery in P2P networks using collaborative agents;the probabilistic double token ring;an efficie...
ISBN:
(纸本)9780889868205
the proceedings contain 40 papers. the topics discussed include: a classified cluster based architecture for resource discovery in P2P networks using collaborative agents;the probabilistic double token ring;an efficient inter-domain routing in wireless mesh networks;a reliable transmission protocol for sensors in poorly connected areas over mobile networks;grid scheduling divisible load with load adaptive computing power;an energy efficient SMT processor with heterogeneous instruction set architectures;a proactive fault tolerance framework for high-performance computing;coverage overlay backfilling: efficient job allocation scheduling scheme for large-scale mesh-connected supercomputers;development of a network on chip for parallel processing systems;a low-power fault-tolerant NOC using error correction and detection codes;link state aware routing for multiple mobile robots in wireless sensor network;and channel assignment protocol with weaker restrictions in wireless multihop networks.
the classification and selection of services within distributed, heterogeneous environments is a non trivial task. For a proper selection and composition of services in such environments - for example in a Grid or Clo...
详细信息
ISBN:
(纸本)9780889868649
the classification and selection of services within distributed, heterogeneous environments is a non trivial task. For a proper selection and composition of services in such environments - for example in a Grid or Cloud - it is required to dispose of detailed information about the existing resources and their characteristics. Particularly for applying appropriate optimization strategies these metadata are indispensable in order to decide which is the actual best resource for a given task, respectively query. It is thus a crucial task to first identify, classify and describe all system components and their attributes in a common language. therefore a generic, high-level model of a service ontology is proposed, that exposes the fundamental building blocks of a distributed, service-oriented environment, by providing a structured collection of components and their functional and non-functional attributes. Finally, the applicability of the proposed ontology is verified by applying it to two distinct application scenarios, one from the domain of distributed Database Management Systems, and one from a High-Energy Physics experiment.
For an initial study in divisible load scheduling, an optimal computing power allocation problem in a distributedparallelcomputing grid involving two sources and a sink is considered. the objective is to optimally a...
详细信息
ISBN:
(纸本)9780889868205
For an initial study in divisible load scheduling, an optimal computing power allocation problem in a distributedparallelcomputing grid involving two sources and a sink is considered. the objective is to optimally allocate the computing power of the sink in the grid in a such way that the total parallelcomputing finish time of the entire load is equalized to the sequential computing finish time while utilizing the full computing power. A numerical method to calculate the optimal adaptive computing power via a deterministic analysis is presented under several computing constraints. Performance of the computing power adaptation is modeled and evaluated. For performance evaluation, we define average computing finish time.
As high-performance computing (HPC) systems continue to increase in scale, their mean-time to interrupt decreases respectively. the current state of practice for fault tolerance (FT) is checkpoint/restart. However, wi...
详细信息
ISBN:
(纸本)9780889868205
As high-performance computing (HPC) systems continue to increase in scale, their mean-time to interrupt decreases respectively. the current state of practice for fault tolerance (FT) is checkpoint/restart. However, with increasing error rates, increasing aggregate memory and not proportionally increasing I/O capabilities, it is becoming less efficient. Proactive FT avoids experiencing failures through preventative measures, such as by migrating application parts away from nodes that are "about to fail". this paper presents a proactive FT framework that performs environmental monitoring, event logging, parallel job monitoring and resource monitoring to analyze HPC system reliability and to perform FT through such preventative actions.
this paper proposes Message Flow Simulator (MFS), which evaluates the communication algorithms for inter- connection network of large-scale parallel computer. MFS calculates communication time from the amount of messa...
详细信息
ISBN:
(纸本)9780889868205
this paper proposes Message Flow Simulator (MFS), which evaluates the communication algorithms for inter- connection network of large-scale parallel computer. MFS calculates communication time from the amount of message flow on communication links. To show the characteristics of MFS, we presented the run time of MFS and the estimated virtual communication time on fat-tree networks including up to 3456 nodes for all-to-all communication. We compared MFS with Booksim 2.0 developed at Stanford University, which is an existing flit-level simulator. the ratio of the throughput of the network estimated by MFS differs from that estimated by Booksim (500 flits/packet) by 2.1% on average (3.6% at maximum and 1.2% at minimum). the simulation results of Booksim were close to the results of MFS when a packet consisted of many flits. We conclude that MFS provides the simulation results that reflect characteristics of the communication algorithms.
Many attempts have been made to optimize the median filter from the software and hardware approach. An architectural design of hardware capable of performing real-time median filtering is presented. the architecture u...
详细信息
ISBN:
(纸本)9780889868205
Many attempts have been made to optimize the median filter from the software and hardware approach. An architectural design of hardware capable of performing real-time median filtering is presented. the architecture uses the histogram approach to calculate the median, while optimizing the sliding window method to reuse all its calculations. Data is output row by row and every input pixel is processed only once. the design is independent of window size or image size, and supports adding more processing elements to support wider images. the control unit design is minimized to enable self-adjustment of plug-and-play processing elements. the architecture is implemented in VHDL and synthesized to a Virtex-2 Pro FPGA. the architecture's performance as well as operation is compared to previous work.
Embedded systems require more and more computational power. Moreover, embedded applications are becoming data-dependent and their execution time depends on their input data. Only a dynamic global scheduling can balanc...
详细信息
ISBN:
(纸本)9780889868205
Embedded systems require more and more computational power. Moreover, embedded applications are becoming data-dependent and their execution time depends on their input data. Only a dynamic global scheduling can balance the workload on the computation resources and reach good performances. thus, a solution to address this problem is to use many-core architectures with a dynamic and centralized control. In this article, we propose new on-line scheduling algorithms adapted to hierarchical many-core embedded systems. the proposed algorithms reduce communications between clusters in order to increase global performance. this paper highlights the good results of a scheduling algorithm named Static Clustering Dynamic Mapping. It consists in dividing the application graph offline and dynamically allocating each part on each cluster.
In two-sided channel routing on a VLSI chip it is often convenient to represent signal nets by trapezoids. In this representation the four corners of the trapezoids are the rightmost and left-most terminals on the upp...
详细信息
ISBN:
(纸本)9780889868205
In two-sided channel routing on a VLSI chip it is often convenient to represent signal nets by trapezoids. In this representation the four corners of the trapezoids are the rightmost and left-most terminals on the upper side and lower side of the channel respectively. the maximum set of nonintersecting trapezoids is of particular interest since corresponding signal nets can be safely assigned to the same layer in the channel routing. Although a sequential algorithm to compute maximum independent set of trapezoids is known, the sweep line approach employed by the sequential algorithm is incremental in nature and does not yield itself to a parallel solution. In this paper we use three new ideas to find the maximum independent set in parallel. First, for every comparable pair of trapezoids we introduce a new unique 'in-between' trapezoid. Next, the trapezoids are mapped to their canonical box representation, and finally, a new parallel operation called 'corner stitching' is applied on boxes to construct chains of boxes which define the independent set. the algorithm presented here is deterministic and is designed to run on a Concurrent Read Concurrent Write parallel random access machine(CRCW-PRAM). the algorithm runs in O(log n) time with O(n2) processors.
Network fault management systems rely heavily on observed alarms to identify the root causes of network failures. Due to the increasing complexity of modern computer networks, the information carried by these alarms m...
详细信息
ISBN:
(纸本)9780889868205
Network fault management systems rely heavily on observed alarms to identify the root causes of network failures. Due to the increasing complexity of modern computer networks, the information carried by these alarms may in fact be vague, imprecise, and inconsistent. thus, these alarms often possess different diagnostic capabilities and should not be treated equally. In this paper, we propose a new distributed alarm correlation approach that effectively tackles the aforementioned data deficiencies. According to the proposed approach, the managed network is first divided into a disjoint set of management domains and each domain is assigned an intelligent agent. Within the framework of Dempster-Shafer evidence theory, the intelligent agent perceives each network entity in its domain as a source of information. As such, alarms emitted by these entities are expected to exhibit different information qualities and are assigned different weights accordingly. Based on their weights, the observed alarms are then correlated by their respective agent into a single local fuzzy composite alarm. Since local composite alarms constitute only partial views of the managed network, they are correlated, by a higher management entity, into a global alarm that accurately reflects a comprehensive view of the managed network.
High Performance computing has been very useful to researchers in the Bioinformatics, Medical and related fields. the bioinformatics domain is rich in applications that require extracting useful information from very ...
详细信息
ISBN:
(纸本)9780889868205
High Performance computing has been very useful to researchers in the Bioinformatics, Medical and related fields. the bioinformatics domain is rich in applications that require extracting useful information from very large and continuously growing sequence of databases. Automated techniques such as DNA sequencers, DNA microarrays & others are continually growing the dataset that is stored in large public databases such as GenBank and Protein DataBank. Most methods used for analyzing genetic/protein data have been found to be extremely computationally intensive, providing motivation for the use of powerful computers or systems with high throughput characteristics. In this paper, we provide a case study for one such bioinformatics application called BLAT running in a high performance computing environment. We use sequences gathered from researchers and parallelize the runs to study the performance characteristics under three different query and data partitioning models. this research highlights the need to carefully develop a parallel model with energy awareness in mind, based on our understanding of the application and then appropriately designing a parallel model that works well for the specific application and domain. We found that the BLAT program is highly parallelizable and a high degree of speedup is achievable. the experiments suggest that the speed up depends on model used for query and database segmentation.
暂无评论