Implementation of processor cells for array algorithms on FPGAs

Abstract

Recent advances in FPGA technology offer a suitable environment for massively parallel, fine-grain array architectures. The paper gives geometric criteria for an optimal "jigsaw tessellated" processor cell, and cost function for cell placement. The paper demonstrates the use of FPGA-based processor arrays by the implementation results of two cellular image processing algorithms. The outlined concepts are being implemented in a placement-routing tool.