Assignment 4

CS 265/571 Advanced Programming Tools and Techniques
Instructors: Jeremy Johnson and Kurt Schmidt
Due date: Fri. Mar. 11 at 11:55 pm

Introduction

The purpose of this assignment is for students to learn how to empirically analyze the runtime of a program and to relate this to the theoretical runtime. Students will also learn how to profile and time programs and to use this information to determine whether to optimize a program and if so what to optimize. Students should review and think about the material in Lecture 8 (chapter 7 of the text). In particular, you will need to use the unix time command, the clock function, and the profiler gprof.

1. Time the C and C++ versions of the markov program on the psalms input using the unix time command. You should redirect output to /dev/null. How do they compare? How do the times compare to what is given in the book? What conclusions can you draw.
2. Profile, using gprof, the C and C++ version of the Markov program. Summarize your findings.
3. Write a main program to time n inserts to the STL map (for different values of n). You should use a map whose index is a string and whose value is an int. You can use strings that are the ascii version of numbers. I.e. you can count from 1 to n and convert each integer to a string and use converted string to index into the map. This way you can easily generate as many strings as you need without reading from a file.
   Use the function clock() to measure the times as was shown in the sample programs (sorti and sortr) to obtain times for different numbers of insertions. You may need to insert a large number of words so that the insertions into map take time larger than the resolution of clock. Plot the times and compute the ratio of the times to n, nlog(n), n^2. What can you say about the growth rate? Can you guess the data structure that is used to implement the map?

What to hand in

Place your solution in a directory called A4. Your directory should contain a file called README that describes all of the files in A4 and summarizes what you did. Make sure to indicate if any of the problems were not done or are not working.

1. README - description of what you did (include the machine name and type with CPU speed, amount of memory, and operating system) and your conclusions.
2. times - a file containing the output of the time command from timing the C and C++ versions of the Markov program running with psalms.txt as input. In addition to the times, you should briefly summarize your results.
3. markov_C.prof - a file containing the output of gprof from running the C version of the Markov program on psalms.txt. Simply redirect the output of gprof to the file.
4. markov_Cpp.prof - a file containing the output of gprof from running the C++ version of the Markov program on psalms.txt. Simply redirect the output of gprof to the file.
5. prof - a file containing your summary of the two profiles of the C and C++ Markov programs.
6. timemap.cpp - C++ source file for the timing insertions into map.
7. maptimes - a file containing pairs (separated by spaces - one pair per line) of input sizes (number of insertions) and times.
8. maptimes.{jpg,ps,eps} plot of the times in maptimes.
9. maptimes.summary - your observations and conclusions regarding the times for map.

Create a gzipped tar file containing a directory called A4 (use the command tar to do this) with your solutions.   If you created a directory called A4 with all of the files for your assignment 2 solution, you can use the command:

$ tar -zcvf A4.tar.gz A4

to create the desired gzipped tar file (the convention is to use the .tar extension for tar files and the .gz extension for gzipped files).  See man tar of info tar for more information about tar.

All assignments should be submitted using webct. You should submit your gzipped tar file.