Scholarlink

The challenges involved in executin 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 (hw) or software accelerators can deliver fast and efficient computation of the NNs, while flexibility can be exploited to support long-term adaptivity. Nonetheless, handcrafting a NN for a specific device, despite the possibility of leading to an optimal solution, takes time and experience, and that’s why frameworks for hw accelerators are being developed. This work, starting from a preliminary semi-integrated ONNX-to-hardware toolchain [23], focuses on enabling Approximate Computing (AC) 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.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

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

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Federico Manca,
  • Francesco Ratto,
  • Francesca Palumbo

摘要

The challenges involved in executin 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 (hw) or software accelerators can deliver fast and efficient computation of the NNs, while flexibility can be exploited to support long-term adaptivity. Nonetheless, handcrafting a NN for a specific device, despite the possibility of leading to an optimal solution, takes time and experience, and that’s why frameworks for hw accelerators are being developed. This work, starting from a preliminary semi-integrated ONNX-to-hardware toolchain [23], focuses on enabling Approximate Computing (AC) 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.