Summary of Optimizing Dnn Inference on Multi-accelerator Socs at Training-time, by Matteo Risso et al.
Optimizing DNN Inference on Multi-Accelerator SoCs at Training-time
by Matteo Risso, Alessio Burrello, Daniele Jahier Pagliari
First submitted to arxiv on: 27 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents ODiMO, a hardware-aware tool that efficiently maps Deep Neural Networks (DNNs) onto heterogeneous Systems-on-Chips (SoCs) during the training phase. The goal is to balance energy consumption or latency with accuracy by strategically splitting individual layers and executing them in parallel on multiple available computing units (CUs). The authors test ODiMO on CIFAR-10, CIFAR-100, and ImageNet datasets using two open-source SoCs, DIANA and Darkside. They achieve a rich collection of Pareto-optimal networks, reducing latency by up to 8x at iso-accuracy on the Darkside SoC and producing up to 50.8x more efficient mappings with minimal accuracy drop. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Offloading Deep Neural Networks (DNNs) to edge devices requires low latency and minimal power consumption. This paper introduces ODiMO, a tool that maps DNNs onto heterogeneous Systems-on-Chips (SoCs) during training. By splitting layers and executing them in parallel on multiple computing units, ODiMO balances accuracy with energy or latency. The authors test this approach on popular datasets like CIFAR-10, 100, and ImageNet using two SoC options. |
