Reuse and Blend: Energy-Efficient Optical Neural Network Enabled by Weight Sharing
Abstract: Optical neural networks (ONN) based on micro-ring resonators (MRR) have emerged as a promising alternative to significantly accelerating the massive matrix-vector multiplication (MVM) operations in AI applications. However, the limited scale of MRR arrays presents a challenge for AI acceleration. The disparity between the small MRR arrays and the large weight matrices in AI necessitates extensive MRR writings, including reprogramming and calibration, resulting in considerable latency and energy overheads. To address this problem, we propose a novel design methodology to lessen the need for frequent weight reloading. Specifically, we propose a reuse and blend (R&B) architecture to support efficient layer-wise and block-wise weight sharing, which allows weights to be reused several times between layers/blocks. Experimental results demonstrate the R&B system can maintain comparable accuracy with 69% energy savings and 57% latency improvement. These results highlight the promise of the R&B to enable the efficient deployment of advanced deep learning models on photonic accelerators.
- Deep learning with coherent nanophotonic circuits. Nature photonics, 11(7):441–446, 2017.
- Prospects and applications of on-chip lasers. eLight, 3(1):1–25, 2023.
- Large-scale photonic chiplet taichi empowers 160-tops/w artificial general intelligence. Science, 384(6692):202–209, 2024.
- Parallel convolutional processing using an integrated photonic tensor core. Nature, 589(7840):52–58, 2021.
- Optical patching scheme for optical convolutional neural networks based on wavelength-division multiplexing and optical delay lines. Optics letters, 45(13):3689–3692, 2020.
- A winograd-based integrated photonics accelerator for convolutional neural networks. IEEE Journal of Selected Topics in Quantum Electronics, 26(1):1–12, 2019.
- Squeezelight: A multi-operand ring-based optical neural network with cross-layer scalability. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 42(3):807–819, 2022.
- Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016.
- 11 tops photonic convolutional accelerator for optical neural networks. Nature, 589(7840):44–51, 2021.
- Neuromorphic photonic networks using silicon photonic weight banks. Scientific reports, 7(1):7430, 2017.
- A high-speed, tunable silicon photonic ring modulator integrated with ultra-efficient active wavelength control. Opt. Express, 22(10):12628–12633, 2014.
- Digital electronics and analog photonics for convolutional neural networks (deap-cnns). IEEE Journal of Selected Topics in Quantum Electronics, 26(1):1–13, 2020.
- Silicon photonics codesign for deep learning. Proceedings of the IEEE, 108(8):1261–1282, 2020.
- On-chip optical convolutional neural networks. arXiv preprint arXiv:1808.03303, 2018.
- High-speed iii-v nanowire photodetector monolithically integrated on si. Nature communications, 11(1):4565, 2020.
- Si microring resonator crossbar array for on-chip inference and training of the optical neural network. Acs Photonics, 9(8):2614–2622, 2022.
- High-speed silicon photonic mach–zehnder modulator at 2 μm𝜇𝑚\mu mitalic_μ italic_m. Photonics Research, 9(4):535, 2021.
- Brecq: Pushing the limit of post-training quantization by block reconstruction. arXiv preprint arXiv:2102.05426, 2021.
- Crosslight: A cross-layer optimized silicon photonic neural network accelerator. In 2021 58th ACM/IEEE Design Automation Conference (DAC), pages 1069–1074, 2021.
- Holylight: A nanophotonic accelerator for deep learning in data centers. In 2019 Design, Automation & Test in Europe Conference & Exhibition (DATE), pages 1483–1488, 2019.
- Libra: Thermal and process variation aware reliability management in photonic networks-on-chip. IEEE Transactions on Multi-Scale Computing Systems, 4(4):758–772, 2018.
- Mitigating the energy impacts of vbti aging in photonic networks-on-chip architectures with multilevel signaling. In 2018 Ninth International Green and Sustainable Computing Conference (IGSC), pages 1–7, 2018.
Paper Prompts
Sign up for free to create and run prompts on this paper.
