I was able to successfully repeat the experiments of the paper.
The scripts and code provided by the authors worked flawlessly 
while the results obtained matched the ones claimed in the paper.

The only differences observed in my results 
were due to the slightly faster hardware in my set-up.
Essentially, all results scaled nicely showing similar behavior as in the paper
and thus not only we can say that the experiments are repeatable but they are
also valid on slightly different hardware as well.


The only minor negative point that one could say about this submission 
is that presentation could be improved.
The authors provided a well organized README file
with detailed steps on how to repeat the experiments for each specific 
graph and table in the paper. In addition, they provided one or two scripts
per graph that would automatically repeat the complete relevant experiment.
For each experiment, the results were collected in 1 or more data files.


Things to improve and make the evaluation much easier would be the following:
Provide an additional single script that performs all experiments in one go.
Results should be collected in exactly the form that they are plotted, 
including x points and preferably collected in a more clean way, i.e., in a single
directory marked as DataFileFigureX for each graph, and each data file should
contain purely the graph data.