In this paper, we study consistent synchronous multi-rate data flow graphs
to determine the minimal required buffer sizes that still guarantee the con
struction of a deadlock-free static schedule. A graph is split up in chains
and clusters that can be studied independently. We present the results for
chains, as well as for the most frequent clusters. The results will be use
d in the rapid prototyping environment GRAPE-II in case the emulation hardw
are contains FPGAs, or when memory is critical.