GD 2017
GD 2016
GD 2015
full list
steering committee
advisory board
organize GD
Literature
books
proceedings
eprints
primer
specification
GraphArchive.org
House of Graphs
benchmarks

It is intended to provide benchmark data for various classes of
graph drawing algorithms. Below are some widely used, publically
available data sets, a more comprehensive collection can be found at
GraphArchive.org.
The newest addition are
 DAGmar level graphs (12 MB)
The DAGmar generator produces DAGs and level graphs. First, it uniformly samples
the leveling, which optionally may be restricted in terms of the number of
levels and the number of vertices per level. In a second step it draws the graph
uniformly from the set of all connected graphs with this leveling. DAGmar also
supports unleveled labeled graphs (DAGs), properleveled graphs, and random
embeddings, i.e., the order of the vertices on the same level. DAGmar is written
in Java, is platformindependent and has no other dependencies. Bundled with the
generator is a suite of pregenerated level graphs, consisting of ten graphs per
each combination of vertex number in {20, 40, ..., 400} and density in
{1.6, 2.6, ..., 10.6}. The tgzfile also contains a technical report describing
the sampling algorithm and the program usage.
Older benchmark are
 Test suites of directed and undirected graphs from the
GDToolkit homepage
(also known as the Rome Graphs)
 Graph Catalog (also known as the AT&T Graphs)
In 2000, an effort to define
GraphML,
an XMLbased data exchange format for graphs and graph drawings,
was initiated.
The following data sets have been generated from the above collections.
Each archive contains graphs with 10 to 100 nodes (see the included
log file for additional statistics):

