Photonic Accelerator
Photonic accelerators leverage the speed and energy efficiency of light to accelerate computationally intensive tasks, primarily focusing on deep neural network (DNN) inference and training. Current research emphasizes developing efficient architectures, such as those based on Mach-Zehnder interferometers and microring resonators, and optimizing algorithms to mitigate challenges like thermal sensitivity and hardware noise, including exploring sparse architectures and in-situ calibration techniques. This field holds significant promise for improving the performance and energy efficiency of AI applications, particularly in resource-constrained environments like edge computing, and for solving complex problems like partial differential equations.
Papers
HEANA: A Hybrid Time-Amplitude Analog Optical Accelerator with Flexible Dataflows for Energy-Efficient CNN Inference
Sairam Sri Vatsavai, Venkata Sai Praneeth Karempudi, Ishan Thakkar
A Comparative Analysis of Microrings Based Incoherent Photonic GEMM Accelerators
Sairam Sri Vatsavai, Venkata Sai Praneeth Karempudi, Oluwaseun Adewunmi Alo, Ishan Thakkar
Towards Efficient Hyperdimensional Computing Using Photonics
Farbin Fayza, Cansu Demirkiran, Hanning Chen, Che-Kai Liu, Avi Mohan, Hamza Errahmouni, Sanggeon Yun, Mohsen Imani, David Zhang, Darius Bunandar, Ajay Joshi
Mirage: An RNS-Based Photonic Accelerator for DNN Training
Cansu Demirkiran, Guowei Yang, Darius Bunandar, Ajay Joshi