This preview shows page 1. Sign up to view the full content.
Unformatted text preview: p chunks of roughly equal length covering the range from sqrt( max ) + 1 to max , and allocate a thread for each chunk. 3. Each thread uses the sequentially computed "seeds" to mark the numbers in its chunk. 4. The master waits for all threads to finish and collects the unmarked numbers. As part of the assignment, you must provide (a) a README file that tries to explain the performance you see in terms of Amdahl's law, and (b) a speedup curve that plots performance as you increase the number of cores (from 1 - 8). You should try to vary the size of max to guage the impact of program size on parallel performance. Make sure to pick large numbers for max (e.g., in the millions to ensure there's enough work for threads to do)....
View Full Document
- Spring '08