SiPhAI: A Reconfigurable Silicon Photonic Interposer Network for AI Acceleration
We propose an energy-efficient and reconfigurable tree-based silicon photonic interposer network, adaptable to varying AI workloads. Our design is substantiated using an iPronics programmable photonics platform and demonstrates significant improvements in energy efficiency when compared to the state-of-the-art and with a different number of chiplets.
Authors
- Ebadollah Taheri (ORCID: https://orcid.org/0000-0001-5099-5155)
- Sudeep Pasricha (ORCID: https://orcid.org/0000-0002-0846-0066)
- Mahdi Nikdast (ORCID: https://orcid.org/0000-0003-4930-2985)
- Mohammad Amin Mahdian (ORCID: https://orcid.org/0000-0001-8084-1050)
Institutions
- Colorado State University (US)
Publication Details
- Published
- 2024-11-10
- DOI
- https://doi.org/10.1109/ipc60965.2024.10799865
- Citations
- 1
- Primary Topic
- Photonic and Optical Devices
- Type
- article
- Field-Weighted Citation Impact
- 0.21