GCN Inference Acceleration using High-Level Synthesis