Accelerating HDC-CNN Hybrid Models Using Custom Instructions on RISC-V GPUs

2025-11-07

AI Summary

To address the low energy efficiency and poor programmability of HDC-CNN hybrid models on general-purpose architectures, as well as the lack of native Hyperdimensional Computing (HDC) support in RISC-V GPUs, this paper proposes a customized instruction-set extension for RISC-V GPUs. The extension comprises four instruction categories: HDC encoding, similarity computation, vector operations, and memory access. It enables the first efficient and programmable execution of HDC-CNN hybrid workloads on an open-source RISC-V GPU, balancing flexibility and performance. Microbenchmark results demonstrate up to 56.2× speedup for the extended instructions. End-to-end inference achieves significantly higher energy efficiency and throughput compared to conventional approaches. This work overcomes the flexibility bottleneck of domain-specific accelerators and establishes a new paradigm for high-energy-efficiency hybrid AI computing at the edge.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Hardware-aware ML
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

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.

Problem

• High energy consumption in neural network training and inference
• Hybrid HDC-CNN models suffer from poor generalizability and programmability
• Limited adoption of efficient hybrid accelerators for complex visual tasks

Innovation

• Custom GPU instructions for HDC operations
• Hybrid HDC-CNN models on RISC-V GPUs
• Four custom instructions boost performance 56x