SIPHON: Silicon Photonic Computing based Chiplet Accelerator for Deep Learning

Chengpeng Xia; Haibo Zhang; Yawen Chen; Amanda S. Barnard

doi:10.1109/JETCAS.2025.3619942

SIPHON: Silicon Photonic Computing based Chiplet Accelerator for Deep Learning

Chengpeng Xia, Haibo Zhang, Yawen Chen Amanda S. Barnard

IEEE journal on emerging and selected topics in circuits and systems, 3619942

09/10/2025

: https://doi.org/10.1109/JETCAS.2025.3619942

https://hdl.handle.net/10523/48412

Chiplet

Neural networks accelerator

Photonic communication

Photonic computing

Photonic interconnect

With the substantial increase in computing workload for deep learning applications, traditional electronic accelerators are facing growing constraints and approaching their practical limits. Silicon photonics has emerged as a promising technology for both communication and computation in accelerating deep learning workloads. Existing photonic accelerators focus on either designing monolithic photonic computing cores or using photonic interconnects with electronic cores for deep neural network (DNN) acceleration. However, the integration of photonic computing within many-core photonic interconnect architectures has not been extensively studied. In this paper, we propose a novel scalable chiplet-based photonic accelerator named SIPHON that leverages both photonic computing and communication for ultrafast and energy-efficient DNN training and inference. A photonic interconnection with dynamically configurable multiple communication modes is proposed to address the broad wavelength and bandwidth demands of general photonic computing cores. We design a photonic computing unit (PCU) for the multiply-accumulate operations and gradient computations in forward and backward propagations. A dataflow is developed to facilitate efficient data reuse and parallel computing by leveraging multiple communication modes. To validate SIPHON's photonic computing, we prototype the optical platform using FPGA, RF, and photonic devices. Simulations on five deep learning models show that, compared with the GPU and the state-of-the-art optical-memristor-based backpropagation accelerators, SIPHON achieves up to 11.5× and 2.2× acceleration in time and 55.4× and 6.0× in energy efficiency for DNN training.

: 9926780551901891
: SIPHON: Silicon Photonic Computing based Chiplet Accelerator for Deep Learning
: Chengpeng Xia
Haibo Zhang
Yawen Chen
Amanda S. Barnard
: IEEE journal on emerging and selected topics in circuits and systems, 3619942
: School of Computing
: IEEE
: Pawsey Supercomputing Research Centre
: 09/10/2025
: The published version is not available in full-text in OUR Archive. Where available, a link to the published version is provided (check the DOI and/or the Files and links section). The full-text item may be open access on the publisher's website. An earlier version of the work (such as authors' accepted manuscript following peer-review or unreviewed preprint/author's original version) may be available in the Files and links section of this record. Alternatively, readers may have subscription access to the full-text from the publisher.
: English
: Journal article