DOCTOR: Dynamic On-Chip Remediation Against Temporally-Drifting Thermal Variations Toward Self-Corrected Photonic Tensor Accelerators

arXiv:2403.02688v1 Announce Type: cross
Abstract: Photonic computing has emerged as a promising solution for accelerating computation-intensive artificial intelligence (AI) workloads, offering unparalleled speed and energy efficiency, especially in resource-limited, latency-sensitive edge computing environments. However, the deployment of analog photonic tensor accelerators encounters reliability challenges due to hardware noises and environmental variations. While off-chip noise-aware training and on-chip training have been proposed to enhance the variation tolerance of optical neural accelerators with moderate, static noises, we observe a notable performance degradation …

abstract accelerators analog artificial artificial intelligence arxiv challenges chip computation computing cs.ai cs.et cs.lg deployment doctor dynamic edge edge computing efficiency energy energy efficiency environments hardware intelligence latency photonic computing reliability solution speed tensor type workloads