Full Index for Basic foilset 
| This module describes many current approaches including different languages which support message passing, data parallelism and task parallelism. We describe the status of various approaches and what software is appropriate for what problems and what machines |
| We describe High Performance Fortran and what features are needed for what applications as well as |
| Special needs of coarse grain task parallelism |
)(focus style:
) Denote Foils where Image Critical
)(focus style:
) Denote Foils where Image has important information
)(focus style:
) Denote Foils where HTML is sufficient
What software is suitable for what problems?
What Applications have we learnt from ?
Comparison of 3 different Programming Models
What software systems are appropriate for what problem architectures -- I?
What software systems are appropriate for what problem architectures -- II?
Candidate Software Paradigms for each problem architecture
Problem v. Machine Architecture
Software Built on Top of FORTRAN, C ...
Evaluation of High Performance Fortran What applications need what features of HPF and its extensions ?
What Issues should High Performance Fortran (HPF) Address!
Goal of High Performance Fortran
Any Complete Programming Environment Must Handle
HIGH PERFORMANCE FORTRAN COMPILERS
What type of compiler is HPF ?
The High Performance Fortran Library
HPF Intrinsic Library
High Performance Fortran Library -- I
High Performance Fortran Library -- II
Fortran 90 Local Routine Intrinsics
Imprecise Mapping of Problem Classes into Runtime and Language Terms
General Applicability of HPF, HPF++, HPC++
Importance of HPF, HPC++ to Users
What about other languages ?
What applications does HPF support? If not - what extensions are needed?
5 Categories of Problems
HPF+: Extensions to HPF -- Use name HPF+ so don't predjudice "official" HPF2
Original Classification used in Planning CRPC (Maryland Rice Syracuse) HPF extensions
What can current HPF Language Surely do?
HPF can also do the synchronous
Current HPF can also do the Embarassingly Parallel
Difficult but (almost) possible for HPF
HPF can express Region Growing in Image Processing
HPF can also express irregular domains seen near critical points of physical systems
Swendsen-Wang clusters (boundaries shown in black) for 3 state Potts model at Tc
Significant improvement in HPF needed but seems possible for Particle in the Cell
Significant improvement in HPF needed but seems possible for .. (contd)
Some very hard Loosely Synchronous Problems -- HPF Expression uncertain
Large N-Body Calculations (Quinn, Salmon, Warren)
10,000 Body Barnes-Hut Tree
8M bodies - 10 Mpc diameter Final state with ~700 resolved "galaxies" (Warren, Quinn, Zurek)
The Largest "Galaxy" Halo with 137,000 particles taken from 8.8 Million particle simulation of Warren,Fullagar,Quinn and Zurek
Speed Up on nCUBE of Parallel Barnes Hut Algorithm
Final Summary of Problem and Software Architectures
What determines when Parallelism is Clear ?
The map of Problem ---> Computer is performed in two or more statges
The Mapping of Space of Problem Architectures onto Space of Machine Architectures
We can divide problems and machines into interconnected modules - and we do this at different granularities - I
We can divide problems and machines into interconnected modules - and we do this at different granularities - II
Different Grain Sizes for MetaProblems and Interpreter
Software Integration -- Support of Coarse Grain Tasks (Metaproblems) using AVS
AVS as System Integration Tool
A distributed parallel computing environment using AVS module network
Case Studies in Integrating AVS into HPDC Applications -- Stock Option Pricing
System Integration and Data Flows for financial modeling on a mix of Workstations, CM5 and Maspar
Option Price Modeling Screen Dump
Electromagnetic Simulation using AVS Screen Dump
Case Studies in Integrating AVS into HPDC Applications -- Electromagnetic Simulation
Physical Problem and Domains studied in Electromagnetic Simulation
Mapping of Electromagnetic Simulation onto MetaComputer
Setup for Computational Electromagnetic AVS Simulation
Data Assimilation -- NASA Grand Challenge Kalman Filters to combine weather models and data
AVS Distributed Computing Setup for Data Assimilation
Summary of Outstanding Issues in Programming Paradigms
Integrating Role of ANDF and possible HPANDF
Some Issues in Programming Paradigms
Some Different Approaches for Software Coordination
Questions in Comparison of AVS and PVM
Software Integration Questions? Full WebWisdom URL and this Foilset Search
This contains all WebWisdom links preceded by those referenced in this foilset