Class 1: More on Parallel Algorithm Design

Announcements, etc. :

1. Remember that the next draft of your project document is due on Thursday.
2. Thursday's class will meet in the Educational Cluster room (GC 297). We will use this class to talk about logistic issues-logging in into the various machines, using Unix machines, LaTeX, LaTeX2html, etc. Please make sure that your Ed Cluster account is functional before then.

In today's class, we will look at the next two phases of algorithm design specified in Chapter 2 of the Foster text.

The third stage of the parallel algorithm design phase is Agglomeration. This stage moves closer to the machine on which the problem will be solved, and involves the grouping together of tasks formed in the Partitioning and Communication stages discussed earlier.

The final stage of the design phase is Mapping. This is the most ``concrete'' of the four stages and involves the mapping of the agglomerated tasks onto the actual processors of the machine.

The details presented in Chapter 2 of the text about Agglomeration and Mapping will be discussed.

Reading: Go through the three case studies at the end of Chapter 2 of the text. There might be similarities with your project problem that would help you think about matters.