Accelerating HDC-CNN Hybrid Models Using Custom Instructions on RISC-V GPUs
Wakuto Matsumi, Riaz-Ul-Haque Mian
TL;DR
Hyperdimensional Computing offers low-power, highly parallel processing but struggles with accuracy on complex tasks, motivating HDC-CNN hybrids. This paper introduces a RISC-V-based GPU accelerator with four custom HDC instructions to accelerate Bound-like operations and integrate seamlessly with CNNs. In microbenchmarks, the custom instructions provide up to 56x speedup, but for real-world image classification gains are modest due to encoding-dominated workloads, highlighting encoding as the bottleneck. The work demonstrates the viability of open, programmable RISC-V GPUs for domain-specific accelerators and points to encoding and matrix-operation optimizations as key future directions.
Abstract
Machine learning based on neural networks has advanced rapidly, but the high energy consumption required for training and inference remains a major challenge. Hyperdimensional Computing (HDC) offers a lightweight, brain-inspired alternative that enables high parallelism but often suffers from lower accuracy on complex visual tasks. To overcome this, hybrid accelerators combining HDC and Convolutional Neural Networks (CNNs) have been proposed, though their adoption is limited by poor generalizability and programmability. The rise of open-source RISC-V architectures has created new opportunities for domain-specific GPU design. Unlike traditional proprietary GPUs, emerging RISC-V-based GPUs provide flexible, programmable platforms suitable for custom computation models such as HDC. In this study, we design and implement custom GPU instructions optimized for HDC operations, enabling efficient processing for hybrid HDC-CNN workloads. Experimental results using four types of custom HDC instructions show a performance improvement of up to 56.2 times in microbenchmark tests, demonstrating the potential of RISC-V GPUs for energy-efficient, high-performance computing.