OpenCL-based deeply pipelined HLS implementation for iterative graph applications

Official URL: https://dx.doi.org/10.1145/3787109.3815260

Graph applications play a central role in different domains such as machine learning, data analytics, natural language processing, and fraud detection. Generating highĝ€'performance kernels tailored to custom accelerator architectures for such workloads remains challenging due to their irregular memory access patterns and dataĝ€'dependent control flow. We propose an OpenCLĝ€'based framework for the automated generation of deeply pipelined High Level Synthesis (HLS) implementations of iterative graph algorithms. The framework integrates a set of optimization techniques that restructure iterative graph algorithms to maximize pipeline utilization, reduce memory bottlenecks, and enable aggressive HLS optimizations. Although broadly applicable to a wide class of iterative graph workloads, we demonstrate the approach using PageRank as a representative case study. The experiment results demonstrate that efficient synthesis and high throughput can be obtained. The generated deeply pipelined HLS kernels can deliver substantial performance benefits for graphĝ€'centric FPGA acceleration.