PipeJSON: Parsing JSON at Line Speed on FPGAs

JavaScript Object Notation (JSON) gained popularity as a data exchange and storage format. While recent advances on modern CPUs show an improved JSON parsing by using data parallelism with vector instructions, the rigid instruction set and limited pipelining of CPUs prevent parsing performance from reaching the practical limit of memory bandwidth. We present PipeJSON, the first standard-compliant JSON parser to process tens of gigabytes of data per second. It utilizes FPGA hardware to make extensive use of pipelining and can parse multiple characters per clock cycle. To ensure usability in software projects, PipeJSON is written in Data Parallel C++ and achieves 7.95x speedup over state-of-the-art JSON parsers on CPU, despite data transfer to the FPGA.