arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2406.09078cs.AR
收藏

面向 FPGA 自适应神经网络推理的 ONNX 到硬件设计流程

ONNX-to-Hardware Design Flow for Adaptive Neural-Network Inference on FPGAs

Federico Manca, Francesco Ratto, Francesca Palumbo

AI总结:

本文扩展半集成 ONNX 到硬件工具链,通过近似计算增强 FPGA 上神经网络推理的适应性,以支持边缘端轻量级、高能效的长期部署。

摘要:

在边缘端执行神经网络(NN)所面临的挑战包括提供多样性、灵活性和可持续性。例如,这意味着要以高能效方式支持不断演进的应用和算法。使用硬件或软件加速器可以实现快速、高效的 NN 计算,同时可以利用灵活性来支持长期适应性。尽管如此,为特定设备手工设计 NN 虽然有可能得到最优解,但需要耗费时间和经验,因此人们正在开发面向硬件加速器的框架。本研究从一个初步的半集成 ONNX 到硬件工具链 [21] 出发,重点利用原始工具链的独特能力来支持适应性,从而实现近似计算。其目标是在边缘端的 FPGA 上实现轻量级、可适应的 NN 推理。

英文摘要:

The challenges involved in executing neural networks (NNs) at the edge include providing diversity, flexibility, and sustainability. That implies, for instance, supporting evolving applications and algorithms energy-efficiently. Using hardware or software accelerators can deliver fast and efficient computation of the NNs, while flexibility can be exploited to support long-term adaptivity. Nonetheless, handcrafting an NN for a specific device, despite the possibility of leading to an optimal solution, takes time and experience, and that's why frameworks for hardware accelerators are being developed. This work, starting from a preliminary semi-integrated ONNX-to-hardware toolchain [21], focuses on enabling approximate computing leveraging the distinctive ability of the original toolchain to favor adaptivity. The goal is to allow lightweight adaptable NN inference on FPGAs at the edge.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑