On compile-time evaluation of process partitioning transformations for Kahn process networks
Sjoerd Meijer, Hristo N. Nikolov, Todor P. Stefanov · 2009
Kahn Process Networks is an appealing model of computation for programming and mapping applications onto multi-processor plat-forms. Autonomous processes communicate through unbounded FIFO channels in absence of a global scheduler. We derive Kahn process networks from sequential applications using the pn com-piler, but the derived networks do not necessarily meet the per-formance requirements. Process partitioning transformations can achieve a more balanced network improving the performance re-sults significantly. There are a number of process partitioning trans-formations that can be used, but no hints are given to the designer which transformation should be applied to minimize, for example, the execution time. Therefore, we investigate a compile-time ap-proach for selecting the best transformation candidate and show results on a Xilinx Virtex 2 FPGA and the Cell BE processor.