Workshop: Twelfth International Workshop on Heterogeneous High-Performance Reconfigurable Computing (H2RC 2026)
Authors: Rye Stahle-Smith, Carter Antley, Jason Bakos, and Rasha Karakchi (University of South Carolina, Dept. of Computer Science and Engineering)
Abstract: Modern heterogeneous high-performance computing (HPC) systems increasingly rely on FPGA-based accelerators to improve throughput and energy efficiency for data-intensive applications. However, many streaming workloads remain constrained by byte-level preprocessing and data movement, which can dominate end-to-end execution time. This paper presents BASE (Byte-level Analytics and Statistical Engine), a reusable streaming preprocessing accelerator that integrates statistical feature extraction directly into the FPGA communication path. BASE performs single-pass histogram generation, entropy estimation, variance computation, and byte-transition analysis while data is transferred through the AXI DMA interface, overlapping preprocessing with communication and reducing host-side computation. Using deployment-time FPGA bitstream screening as a representative workload, experiments on the ZCU104 show that BASE reduces preprocessing latency by 5.04× and improves end-to-end execution time by 4.23×. On the U55C, BASE achieves 1.32× preprocessing and 1.78× end-to-end speedups. These results demonstrate BASE’s potential as a reusable building block for streaming analytics in reconfigurable HPC systems.
Back to Twelfth International Workshop on Heterogeneous High-Performance Reconfigurable Computing (H2RC 2026) Archive Listing
Back to Full Workshop Archive Listing