J.Silc, B.Robic.<br>
Program graph partitioning for macro-dataflow.<br>
<i>Proc. ISSM Int'l Workshop on Parallel Computing</i>, pp.198-200, 
Trani, Italy, September 10-13, 1991.<p>

The paper examines the problem of dataflow graph  partitioning aiming to 
improve the efficiency of macro-dataflow  computing on a ring architecture. 
The partitioning consists of dataflow graph synchronization, and scheduling 
of the synchronous graph. A new scheduling algorithm, called Global Arc
Minimization (GAM), is introduced. Finally, the performance of GAM algorithm 
is evaluated  relative to some other known heuristic methods for static 
scheduling of the FFT. When interprocessor communication delays are taken 
into account, it may be concluded that GAM algorithm achieves adequate 
performance of FFT on the ring architecture.