In December 2019, Cisco announced the Silicon One line of processors, designed to revolutionize the Internet economy and foster innovation. Now, they have introduced the fourth generation consisting of the Cisco Silicon One G200 and Cisco Silicon One G202 devices. The G200 model focuses on advanced Ethernet-based artificial intelligence/machine learning (AI/ML) deployments. It is a 5nm SerDes device capable of 51.2 Tbps, 512 x 112 Gbps. With programmability, low latency, and advanced visibility and control features, it’s an excellent choice for large-scale networks. For its part, the Cisco Silicon One G202 offers similar benefits to customers who want to use 50G SerDes to connect optics to the switch. It is a 5nm SerDes device capable of 25.6 Tbps, 512 x 56 Gbps, with similar features to the G200 model but with half the performance.
Cisco Silicon One G200 and G202
Both the Cisco Silicon One G200 and G202 appliances have a unified architecture that optimizes real-world performance for AI/ML workloads, reducing cost and network latency through major innovations. They also enable multiple deployments across customer networks, from routing functions to large-scale front-end and back-end networks, significantly reducing operational timelines and costs. The Silicon One line condenses more than 2,300 elements onto a single chip, significantly reducing considerably the size of the chassis and improving energy efficiency. The G200 model has double the power efficiency and has twice the latency compared to its predecessor, the Cisco Silicon One G100 device, which was already optimized.