Photonic Systolic Array for All-Optical Matrix–Matrix Multiplication

Download:

• Paper
• Notebook

Abstract:

Systolic arrays have proven to be highly efficient for parallelized matrix–matrix multiplication (MMM), utilizing synchronized, heartbeat-like data flows across an array of processing elements. While optical structures, such as waveguide crossbar arrays and Mach-Zehnder interferometer-based meshes, serve as photonic equivalents to the systolic arrays, the disparity between the two input matrices for multiplication—one using optical signals and the other with system-defined parameters—gives rise to a bottleneck in modern machine-learning tasks, such as evaluating attention scores in large language models. Here, a photonic systolic array that performs MMM entirely with optical signals is proposed, utilizing homodyne detection at each array cell. Adjoint-based design of compact on-chip freeform optical modules enables precise control of light flow without bulky waveguide coupling schemes. The operation of 4×4 and 2×2 photonic systolic arrays are numerically verified, achieving a theoretical computation density of 4.4 PMACs/mm2/s. This design marks a significant step toward practical photonic computing hardware for modern AI workloads.

Citation

J. Kim et al., “Photonic Systolic Array for All-Optical Matrix–Matrix Multiplication.” Laser & Photonics Reviews: e01995 (2025). https://doi.org/10.1002/lpor.202501995.

@article{Kim2025_systolic,
    title = {Photonic Systolic Array for All‐Optical Matrix–Matrix Multiplication},
    ISSN = {1863-8899},
    url = {http://dx.doi.org/10.1002/lpor.202501995},
    DOI = {10.1002/lpor.202501995},
    page = {e01995},
    journal = {Laser \& Photonics Reviews},
    publisher = {Wiley},
    author = {Kim,  Jungmin and Zhou,  Qingyi and Yu,  Zongfu},
    year = {2025},
    month = nov 
}

Related material