Brief Introduction.
We have been guided by the premise that reproducibility is a defining attribute
of empirical research. So far, there have been two generations reporting the
XMT empirical work. The first generation includes the SPAA'98, and WAE'99
papers. The second generation includes the MTEAC'2000, HIPS'01 and TOCS'03
papers.
The first generation of XMT empirical work. A very high-level
description of several algorithms appears in the SPAA98 paper. The following
four documents provide raw data that, with the addition of minimal analysis of
algorithm execution, should enable reproduction of the numerical results
reported in the SPAA98 paper. The WAE99 publication addressed list ranking--the
most involved problem considered in these four documents--and demonstrated how
to incorporate this added analysis. It improved the readability of this raw
data, providing a detailed high level description of the algorithm, and
reviewed the programming techniques used. The WAE99 paper presented algorithms
and code design decisions and optimizations made, comparing them with other
alternatives. Where needed, further elaboration on how to derive the numerical
results from the raw data were given. It also explained why compilation is
feasible.
Note: All documents below are in PostScript. The gzip compressed files can
be viewed directly by using ghostview.
gzipped PostScript (359 KB), PostScript (1.4 MB)
gzipped PostScript (58 KB), PostScript (230 KB)
gzipped PostScript
(81 KB), PostScript (337 KB)
gzipped PostScript (1.1 MB), PostScript (4.1 MB)
The second generation of XMT empirical work. The results reported in the
second generation papers were obtained using the following XMT tools.
Supplemental Documentation
The following document is Shlomit Dascal's record of specifications and
assumptions for the bridging models and their executions. It is a useful
complement to the SPAA98 paper.
Note: Documentation is in PostScript. The gzip compressed files can be
viewed directly by using ghostview.