CSCI-1200 Computer Science II — Fall 2008
Homework 10 — Performance and Order Notation
In this final assignment for CSII, you will carry out a series of tests on the fundamental data structures in the
Standard Template Library to evaluate the relative performance of these data structures and solidify your
understanding of order notation. Be sure to read the entire handout before beginning your implementation.
The Program
You will measure the runtime and the number of pairwise comparisons needed for a few common, mod-
erately compute-intensive operations: sorting, removing duplicates (without changing the overall order),
and finding the mode. You will perform these tests on STL string objects that can be read from a file
or constructed from random sequences of chars. The data structure, operation, size of the test (number
of strings), source of the input, and output file are specified on the command line. Here are two sample
input lines:
a.exe vector sort 10000 random 5 out.txt
a.exe vector mode 1000 in.txt out2.txt
The first example will generate 10,000 random strings of length 5, use a vector to sort them, and then
output the result to the file named “out.txt”. The second example will read the first 1,000 strings from the
file named “in.txt”, use a vector to find the most frequently occurring value (implemented by first sorting
the data), and then outputs that string (the mode) to a file named “out2.txt”. We provide a starting base
of code that performs these operations using the STL vector data structure.
You will extend this program to allow other data structures in place of vector, including: STL list, STL
set or map, STL priority queue, and the cs2hashset implementation of hash tables. For extra credit
you may test other data structures. You should carefully consider the most efficient way (minimize the
running time) to use the data structure to complete the operation. Similarly, for extra credit you may
test other common operations. In the provided code base, two fixed length arrays of string objects are
used to load and output the data. The data structure specified on the command line is the only additional
data structure that is “allowed” in the implementation of the operation. Thus, some combinations of
data structure and operation may not be feasible. In these cases include a short writeup in your report
explaining why the pairing of data structure and operation is impractical.
Measuring the Performance and # of Comparisons
The provided code demonstrates how the clock() function can be used to measure the processing time
of the computation. The resolution accuracy of the timing mechanism is system and hardware dependent
and may be in seconds, milliseconds, or something else. If the resolution on your system is coarse, you
must base your analysis on the measurements from sufficiently large datasets. The program reports the
time to load, process, and save the data. The provided code also demonstrates how the number of pairwise
comparisons used by the operation can be counted by implementing wrapper functions for the <,>, and
== string comparison functions. Functors are also provided for use with set,map,priority queue, and
cs2hashset. Both measurements should inform your analysis of the data structures.
The Report
You will submit the source code for the program described above, but the bulk of the points for this
assignment will be awarded for a complete and well-written report that presents and analyzes
the results of your testing. The basic outline of the report should include: an introduction with your
initial hypotheses; the procedure which includes any nontrivial implementation details; the data from your
tests with some intermediate analysis; and a conclusion that summarizes your findings about the data
structures, in what instances the different data structures are most useful, and any surprises you found in
the results. Any unanswered questions in your analysis that require further testing, should be described
as future work.
