The following analyses are suggested to be run:

• An analysis using Amdahl’s law and the ratio of serial/parallel code for each method;

• time comparison of pipes vs socketpair IPC and effect of different chunk sizes;

• analysis of total execution time for varying numbers of processes/threads;

• analysis of total resident memory for varying number of processes; and

• comparison of parallelization methods (processes vs threading vs OpenMP) using timing and total resident memory and discuss possible differences in speed, memory, communications delay, operating system overhead, etc.

You should write a bash script called performance analysis.sh that automatically runs the various analyses, collates the results for each parallelization strategy, prints a numerical report in the terminal, and generates plots using the gnuplot program. It should save the gnuplot plots into PNG formatted image files for inclusion into your Powerpoint slides. You are to use /usr/bin/time to do the timing and memory measurements. Also feel free to use any combination of bash, sed and awk to facilitate the analyses (e.g. using bash for loops to cycle through parameters to pass to your program, sed and awk to extract the relevant numbers, etc.).