The zip-file Bench.zip contains 180 completely random task graphs. The names of task graph files are t__.td, where should be substituted with the number of task in that file, with edge existence density, while $i$ is the index of graph with same $n$ and $r$ value (there are 6 graphs for each ($n$, $r$) pair). The graphs in that zip-file does not depend on multiprocessor architecture, they can be scheduled to arbitrary multiprocessor system, but, since optimal solution is not known, the solution quality can not be estimated. In each file, data are written in following format: - first row contains the value for the number of tasks n; - next n rows contain following data: task index i=1,...,n, task duration t_i, i=1,...,n, number of successors n_{succ} and list of n_{succ} pairs s_j c_{ij} representing index of j-th successor and corresponding communication.