Research in Scalable Computing
Outline
Scalable Computing
A Distributed Network-Computing Environment
Basic Approach
Network Environment
Sensitivity of Non-Dedicated Computing - Process migration
Data Collection and Restoration
Memory Space Representation
MSR(cont.)
Timing Results in Seconds
Scalable Networks of Workstations (SNOW)
Components of SNOW
Function of MCL
Software Components
Buffer Data Transfer (BDT) Mechanism
Exection Time Comparison
Major Results
HAPANS
Parallel Algorithms for Compact Schemes
4th Order Direct Solver
Table of Computation Time
Current Work: Wavelet Collocation Method
Advantage of Wavelet ADI Method
Adaptive Wavelet ADI for Flame Propagation
Performance Evaluation
Models of Speedup
Fixed-Size Model (Amdahl Law)
Fixed-Time Model (Gustafson)
Memory-Boundary Model (Sun & Ni)
Model of Speedup
Parallel Performance Metrics (Run-time is the dominant metric)
Scalability
Scalability(cont.)
Isospeed Scalability
The Relation of Scalability and Time
Scaled Crossing Point
Application of Scaled Crossing Point
Range Comparison Via Performance Crossing Point
Range Comparison
Performance Prediction
Automatic Performance Prediction
The SCALA (SCALability Analyzer) System
The Structure of SCALA
Prototype Implementation
3D Demo
Future Research
Acknowledgement
Email: sun@bit.csc.lsu.edu
Home Page: http://www.csc.lsu.edu/~scs
