LightningSim Benchmark Projects

This dataset consists of 33 High-Level Synthesis projects for AMD/Xilinx Vitis HLS 2021.1, which were used in the evaluation of LightningSim: Fast and Accurate Trace-Based Simulation for High-Level Synthesis. These benchmark projects come from several sources: Projects whose names contain “basic_examples_vhls” are derived from Xilinx Vitis HLS Introductory Examples, version 2021.1. Projects whose names contain “pp4fpgas” are derived from the examples from Kastner et al.’s <em>Parallel Programming for FPGAs</em> book. Projects whose names contain “flowgnn” are derived from Sarkar et al.’s “FlowGNN: A Dataflow Architecture for Universal Graph Neural Network Inference via Multi-Queue Streaming.” All projects were modified to ensure that the “setup.tcl” script inside each project directory can be used to setup a Vitis HLS project with name “project” containing a solution with name “solution1” which is able to be synthesized and co-simulated under AMD/Xilinx Vitis HLS 2021.1.