Muutke küpsiste eelistusi

Annual Review Of Scalable Computing, Vol 4 [Pehme köide]

(Nus, S'pore)
  • Formaat: Paperback / softback, 296 pages
  • Sari: Series On Scalable Computing 4
  • Ilmumisaeg: 25-Mar-2002
  • Kirjastus: World Scientific Publishing Co Pte Ltd
  • ISBN-10: 9810249519
  • ISBN-13: 9789810249519
Teised raamatud teemal:
Annual Review Of Scalable Computing, Vol 4Suurem pilt
  • Formaat: Paperback / softback, 296 pages
  • Sari: Series On Scalable Computing 4
  • Ilmumisaeg: 25-Mar-2002
  • Kirjastus: World Scientific Publishing Co Pte Ltd
  • ISBN-10: 9810249519
  • ISBN-13: 9789810249519
Teised raamatud teemal:
Püsilink: https://www.kriso.ee/db/9789810249519.html
A collection of seven long articles, this book comprehensively discusses significant projects in scalable computing in various research organizations around the world. It represents the quantitative and qualitative growth of work in the area.
Experiences with Shared Virtual Memory on System Area Network Clusters: System Simulation, Implementation, and Emulation
1(50)
Introduction
1(2)
Overall Methodology and Results
3(4)
System Simulation
3(1)
System Implementation
4(2)
System Emulation
6(1)
System Simulation
7(10)
Simulation Environment
7(2)
Methodology
9(1)
Applications
10(1)
Effects of Communication Parameters
11(3)
Limitations on Application Performance
14(2)
Degree of Clustering
16(1)
System Implementation
17(9)
Network Interface and SVM Protocol Extensions
18(2)
Experimental Testbed
20(1)
Applications
21(1)
Results
21(3)
Remaining bottlenecks
24(2)
System Emulation
26(10)
Emulation infrastructure
26(4)
Impact of Fast Interconnection Networks
30(2)
Impact of Wide, CC-NUMA Nodes
32(4)
Discussion on Methodology
36(1)
Related Work
37(2)
Conclusions
39(2)
Acknowledgments
41(1)
Bibliography
41(10)
Average-Case Scalability Analysis of Parallel Computations
51(45)
Introduction
51(3)
Motivation
51(1)
Overview of the
Chapter
52(2)
Scalability of Parallel Computations
54(4)
Isoefficiency Scalability
54(2)
Isospeed Scalability
56(1)
Average-Case Scalability
57(1)
Task Precedence Graphs on Multiprocessors
58(15)
A Probabilistic Model
58(1)
Independent Tasks
59(4)
Dependent Tasks
63(1)
Iterative Computations
63(1)
Search Trees
64(3)
Partitioning Algorithms
67(2)
Diamond Dags
69(2)
Summary
71(2)
Task Interaction Graphs on Multicomputers
73(18)
A Probabilistic Model
73(1)
Symmetric Static Networks
74(5)
Completely Connected Networks
79(1)
Rings
80(1)
Meshes with Wraparound Connections (Tori)
81(2)
Hypercubes
83(2)
k-ary d-cubes
85(5)
Comparison and Discussion
90(1)
Summary
91(1)
Further Research
91(2)
Bibliography
93(3)
Parallel IO Prefetching and Caching
96(56)
Introduction
96(11)
Improving I/O Performance with Parallel I/O
97(1)
Overview of Parallel I/O Organizations
98(1)
Prefetching and Caching
99(2)
Simplifying the Problem: Read-Once Accesses
101(2)
Caching: Dealing with Repeated Accesses
103(1)
An Application to Real-Time Video Scheduling
104(2)
Organization of This
Chapter
106(1)
Formalizing the Scheduling Problem
107(2)
Model of Parallel I/O System
107(1)
The Parallel I/O Scheduling Problem
108(1)
Offline Prefetching and Caching
109(8)
Read-Once Scheduling
109(2)
Read-Often Scheduling
111(6)
Online Prefetching and Caching
117(30)
Lookahead
118(1)
Performance Metrics
119(1)
Lower Bounds
120(10)
Read-Once Scheduling
130(11)
Read-Often Scheduling
141(6)
Summary
147(1)
Bibliography
148(4)
A C++/Tuple-Lock Implementation for Distributed Objects
152(30)
Introduction
152(2)
Linda
153(1)
C++/Tuple-Locks
154(1)
Tuple-Lock
154(3)
Using C++
157(1)
Primitives
157(3)
Design and Implementation
160(8)
Parallel Virtual Machine
160(1)
Preprocessor Link
160(1)
Shared Object Handler
161(3)
Central Registrar
164(1)
Distributed Processing
164(1)
Replication
165(3)
Examples
168(7)
One-Way Bridge Problem
169(1)
Jacobi
169(3)
FIFO Queue
172(3)
Results
175(3)
Startup Costs
176(1)
Tuple-Lock Costs
176(2)
Advantages and Disadvantages
178(1)
Conclusion
179(1)
Bibliography
179(3)
Static Data Allocation and Load Balancing Techniques for Heterogeneous Systems
182(43)
Introduction
182(1)
Static Data Allocation Strategies for Linear Arrays
183(10)
A Motivating Example
184(1)
Distributing Independent Chunks
185(3)
Finite-Difference Computations
188(1)
Linear System Solvers
189(4)
Static Data Allocation Strategies for Geometric Problems
193(13)
MM Algorithms
193(3)
The Heterogeneous MM Optimization Problem
196(7)
Related Work
203(3)
Master-Slave Tasking
206(8)
Problem Statement
207(2)
With Initial and Final Communications
209(1)
Solution With an Initial Scattering of Data
210(1)
Solution With Initial and Final Communications
211(3)
Related Work
214(1)
Conclusion
214(1)
Bibliography
215(10)
Building a Global Object Space for Supporting Single System Image on a Cluster
225(36)
Introduction
225(1)
Overview of JESSICA
226(5)
Main Features of JESSICA
227(1)
JESSICA System Architecture
228(2)
Preemtpive Thread Migration
230(1)
Transparent Redirection
231(1)
Design Issues of the Global Object Space
231(7)
The GOS Architecture
231(2)
GOS Initialization and Object Allocation
233(1)
Garbage Collection
234(1)
Criteria for an Efficient GOS
234(4)
Factors Contributing to GOS Efficiency
238(13)
Memory Consistency Models
238(5)
Coherence Protocols
243(6)
Implementation Optimizations
249(2)
Performance Evaluation
251(3)
Related Work
254(1)
Future Work
255(2)
Conclusion
257(1)
Acknowledgements
257(1)
Bibliography
257(4)
A Computation-Centric Multilocation Consistency Model for Shared Memory
261(20)
Introduction
261(3)
Computation-Centric Framework
264(2)
MLC: Characterization
266(3)
Weakening MLC: The MLC Hierarchy
269(1)
Backer is Multilocation Consistent
270(2)
Strengthening MLC: Acquires and Releases
272(1)
Related Work
273(1)
Conclusion
273(2)
Bibliography
275(6)
Proof of Acyclicity Theorem (i)
277(1)
Proof of Reasonableness Theorem
278(2)
Backer
280(1)
Proof of Backer Theorem
281