High-speed FPGA-based Design and Implementation of Text Search Processor

Abstract

In the age of computer evolution, the number of data grows swiftly. Moreover, the requirement of extracting the information from the database becomes urgent. Full-text search provides methods to quickly locate multiple keywords inside extensive text data and has gained more consideration in recent years. The proposed tools, such as Lucene, Hyper Estraier, and Namazu, are based on general-purpose processors. They spend more time on index input documents and require more space to store these indexes. In this work, we provide a text search processor design that could perform the full-text search without indexing. The text search processor offers a high-performance, high-level of parallelism and scalability. The design is deployed on Field Programmable Gate Arrays (FPGA) platforms. More than 70K processing units can be integrated on Xilinx Alveo U50. The working frequency achieves 266-MHz after place and route.