⚡ 本页包含 AI 生成的分析内容,仅供参考
本文提出一款采用5nm FinFET工艺的200GS/s 8位接收机,具有超过60GHz的模拟前端带宽,实现了20fJ/c-s的能效,用于800Gb/s光相干通信。该设计解决了超高速光通信中接收机带宽与功耗的瓶颈问题。
C. Nani4, D. Albano4, F. Ahmad1, F. Solis2, G. Minoia4, G. Hatcher1, M. Bachu3, M. Garampazzi4, M. Hassanpourghadi1, N. Fan1, P. Prabha1, S. Fan3, S. Ho5, T. Dusatko5, T. Wu1, W. Elsharkasy1, Z. Sun6, S. Jantzi1, L. Tse3 Marvell, Irvine, CA Marvell, Cordoba, Argentina 3 Marvell, Santa Clara, CA 4 Marvell, Pavia, Italy 5 Marvell, Vancouver, Canada 6 Marvell, Singapore, Singapore 1 2 The post-COVID era has seen an unprecedented rise in activities requiring high bandwidth consumption such as remote working, streaming, VR, and AR. With the recent breakthroughs in AI and the introduction of Chat GPT, data transport and interconnect fabrics of data centers are being stressed to the limits and forced to scale up even more rapidly than their traditional pace. This is leading to a throughput expansion rate that is faster than existing thermal handling capability while forcing the power form factors of
R. L. Nguyen1, A. Mellati1, A. Fernandez2, A. Iyer3, A. Fan1, B. Reyes2, C. Abidin1,