In order to generate the HTML based comparison sheet, perform the following steps.

0. Obtain input files, I've used the data sets Sinha&Zobel used in their
   Burstsort experiments. You can find them via Google. They are about
   300 megabytes each.

1. Run the ''benchmark'' script, takes about 12 hours if you run all algorithms
   and three input files, each about 300MB. This runs each algorithm seven
   times for each input file.

2. Run OProfile with the ''collect-oprofile-statistics'' script. Takes about 12
   hours. Requires root priviledges.

3. Run memusage with the ''collect-memusage-statistics'' script. Takes about 30
   minutes.

4. Process the OProfile results with ''process-oprofile-statistics'' script.
   Takes about one hour.

5. Process memusages results with ''process-memusage-statistics'' script.
   Takes a minute or two.

6. Finally, run ''make''. This will generate the HTML file based on the results
   of the previous steps.


NOTES:
*) You can create results for smaller subsets of available algorithms using
   environmental variables, for example:
      export ALGS="1 2 3"
   OR:
      ALGS="1 2 3" ./collect-oprofile-statistics
      ALGS="1 2 3" ./collect-memusage-statistics
      etc.