JExplore: Design Space Exploration Tool for Nvidia Jetson Boards

2025-02-16

AI Summary

The vast hardware configuration space of NVIDIA Jetson edge AI platforms and the difficulty of cross-board performance optimization hinder efficient deployment. Method: This paper proposes the first hardware-software co-design space exploration framework supporting arbitrary search algorithms. It standardizes the exploration interface to uniformly abstract communication control, dynamic configuration scheduling, and multi-dimensional performance monitoring—including power consumption, latency, and throughput—thereby significantly lowering the barrier to customized tuning. Implemented in Python/C++, the system integrates SSH-based remote management, Jetson Clocks API, nvtop/SMI monitoring, and is compatible with NSight Systems and custom profilers. Results: On Jetson Orin and AGX Xavier platforms, configuration evaluation speed improves by 5–8×. For representative models—YOLOv5, ResNet50, and DeepLabv3—the framework identifies optimal energy-efficient configurations, achieving up to a 2.3× improvement in energy efficiency.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Algorithm Configuration
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Nvidia Jetson boards are powerful systems for executing artificial intelligence workloads in edge and mobile environments due to their effective GPU hardware and widely supported software stack. In addition to these benefits, Nvidia Jetson boards provide large configurability by giving the user the choice to modify many hardware parameters. This large space of configurability creates the need of searching the optimal configurations based on the user's requirements. In this work, we propose JExplore, a multi-board software and hardware design space exploration tool. JExplore can be integrated with any search tool, hence creating a common benchmarking ground for the search algorithms. Moreover, it accelerates the exploration of user application and Nvidia Jetson configurations for researchers and engineers by encapsulating host-client communication, configuration management, and metric measurement.

Problem

• Optimal configuration search for Nvidia Jetson
• Design space exploration tool
• Host-client communication and configuration management

Innovation

• Multi-board design space exploration
• Integration with search algorithms
• Encapsulates communication and configuration management