Lecture19 - IE 495 Lecture 19 November 2 2000 Reading for...

Info iconThis preview shows pages 1–7. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: IE 495 Lecture 19 November 2, 2000 Reading for This Lecture Primary ¡ Horowitz and Sahni, Chapter 8 ¡ Grama and Kumar, Parallel Search Algorithms... Parallel Branch and Bound Divide and conquer approach "Obvious" approach to parallelization Parallelize recursive version What are the problems with this? How does this compare to other divide and conquer algorithms (such as merge sort)? A Better Approach Master-slave model Master process maintains ¡ a priority queue of nodes ¡ a pool of slave processes to process the nodes Whenever a slave finishes processing a node, the master determines its next course of action ¡ keep one (or more) of the children ¡ get a completely new node Performance Measures Overall running time Measures of overhead/redundant work ¡ Size of search tree ¡ Average time to process a node Measures of idle time ¡ Time slaves spend waiting for work ¡ Percentage load of tree manager Scalability Issues Master process will become a bottleneck...
View Full Document

This note was uploaded on 08/06/2008 for the course IE 495 taught by Professor Linderoth during the Fall '08 term at Lehigh University .

Page1 / 11

Lecture19 - IE 495 Lecture 19 November 2 2000 Reading for...

This preview shows document pages 1 - 7. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online