=========================================================
Comparing Integer Data Structures for 32 and 64-bit Keys

          Nicholas Nash and David Gregg.
=========================================================
               
0. Introduction
1. Compilation
2. Result Generation
3. Paper compilation

---------------
0. Introduction
---------------

This file describes how to compile and generate the results described in the 
paper. It assumes that the compilation and results gathering are being
executed under the BASH shell, and gcc/g++/make, Python, Perl, gnuplot and latex are all 
available. 

All commands given below are described relative to the directory where this file (README) resides
in the extracted archive.

--------------
1. Compilation
--------------

Compilation of the code for the data structures is very simple, simply type:


$ cd expts
$ make


This will build a single binary 'perf_test' that can be used
for comparing the performance of the data structures.

--------------------
2. Result Generation
--------------------

Gathering the results can be accomplished by simply typing:

$ cd expts
$ ./gather.py

This is likely to take a several hours at the minimum.
To have the experiments continue after you log out,
and so that you can monitor them remotely with a 'tail -f',
this might be useful:

$ nohup ./gather.py 1>progress 2>errors &

To average the results into the form used in the paper,
there is a script called 'average.py'.

Simply supply this script with the directory of the results
greater by 'gather.py'.

--------------------
3. Paper Compilation
--------------------

To compile the paper to a pdf, type:

$ cd jea_paper
$ latex jea_paper
$ dvipdf jea_paper.dvi

If you wish to see freshly generated 
plots in the paper type:

$ cd jea_paper/gnuplot_scripts
$ ./gnuplot_dir.pl

Then recompile the paper as just described.

Note that all the data presented in the paper is
present in the directory 'jea_paper/data'

Furthermore, additional results for a Sun UltraSPARC IIIi
architecture may be found in 'jea_paper/data/averaged_melody_results'