Tuesday/Thursday 3:00--4:30

1-390

Prof. Alan Edelman

Course Poster

This year we will use Silicon Graphics, a Dec Turbolaser Server, an IBM SP-2 and perhaps networks of workstations. (No more CM-5). As a new educational experiment, we will have joint lectures with Berkeley using the PictureTel Videoconferencing system in 9-253. Perhaps in the future we will see the following cross-listing MIT 18.337/ UCB CS 267 . Also worth noting are the interesting BU tutorials.

- Cheewee Chew Parallelized Structural Optimization
- Yanyuan Ma Inverse of a Matrix Related To a Graph
- Mats Nigam Parallelization of the Uzawa Algorithm
- Scott Palmtag Parallel Domain Decomposition Solution to the Neutron Diffusion Equation
- David Quiram
- Rajeev Surati Micromechanical Device Direct Simulation Monte Carlo
- Linqing Wen and Matias Zaldarriaga A parallel halo identification code for N-body simulations.

Class Notes (old unfinished version)

Course Syllabus

Problem Sets

Computing Platforms

- BU's 18 Processor SGI Power Challenge Array
- MIT's 12 Processor IBM SP-2
- Digital's 8400 AlphaServer

Parallel Languages

Lectures

- Lecture 1: (2/6) Introduction to Parallel Machines
and Parallel Programming
- Top 500 supercomputers
- The computer vendors
- Class notes: The least you should know about how computers are built
- Class notes: Architecture of Parallel Computers

- Lecture 2: (2/8) Floating Point Arithmetic (Room 9-253)
- Lecture 3: (2/13) Silicon Graphics architecture and environment
- Lecture 4: (2/15) Parallel Prefix and Data Parallelism
- Lecture 5: (2/22) Matrix Multiply on One Processor
Scientific Software Libraries: Machine Single Processor Multiprocessor IBM SP-2 ESSL PESSL Dec 8400 DXML SGI sgimath

- Lecture 6: (2/27) Parallel Matrix Multiply
- Lecture 7: (2/29) Lapack and Scalapack
- Lecture 8: (3/5) High Performance Fortran
- Lecture 9: (3/7) More HPF and Spectral Partitioning
- Lecture 10: (3/12) Fast Multipole Algorithm
- Lecture 11: (3/14) Fast Multipole Algorithm II
- Lecture 12: (3/19) N-body Simulations
- Lecture 13: (3/21) Domain Decomposition
- Lecture 14: (4/2) Geometric Mesh Partitioning
- Lecture 15: (4/4) Climate Models
- Lecture 16: (4/9) Domain Decomposition Part II
- Lecture 17: (4/11)
- Lecture 18: (4/18) Fourier Transform with representation theory
- Lecture 19: (4/23) Guest Lecture John Eck on obtaining performance from symmetric multiprocessors
- Lecture 20: (4/25) Guest Lecture Shang-Hua Teng on Spectral Partitioning Works
- Lecture 21: (4/30) Multipole through Interpolation
- Lecture 22: (5/2) Interpolation Sparse Matrices
- Lecture 23: (5/7) Berkeley Projects
- Lecture 24: (5/9) Non equispaced FFT?
- Lecture 25: (5/14) Wavelets?
- Lecture 26: (5/16) Student Presentations

Documentation

- Matlab Online Reference
- Matlab Primer (postscript)
- Matlab On-line Reference
- GAMS - Guide to Available Math Software

A search facility to find numerical software - Notes on IEEE Floating Point Arithmetic (postscript)

Ancient History

Maintained by Alan Edelman. Last modified: February 8, 1996