Scheduling and allocation using closeness tables

An heuristic approach towards the scheduling, binding and allocation problem for the high-level synthesis of data-paths is presented. The approach makes use of closeness tables to group operations with similar closeness properties, based on their inputs and outputs, into clusters. A tight packing scheduling and binding algorithm is then used to schedule and bind operations from individual clusters to individual functional units. A low interconnect solution is subsequently generated as a result of binding similar groups of operations with common sources and sinks to the same functional units. The approach simultaneously generates efficient schedules. It is shown that this is achieved in fast execution time and that the problem can be solved in reasonably low time complexity. Comparisons are made against other nonpipelined and pipelined approaches.