Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


Xilinx Demonstrates Breakthrough Optical Networking on the road to 7 nm

Today Xilinx announced the FPGA industry’s first demonstration of breakthrough 112G PAM4 electrical signaling technology for optical networks, as well as announcing the addition of 58G PAM4 transceivers to its 16nm Virtex UltraScale+ portfolio.

Xilinx has a long history of driving standards efforts and pushing performance limits in serial interconnect technology, and continues to do so with the industry’s first 112G PAM4 demo as well as with its 58G PAM4 solution, here today for customers to begin designing in with,” said Karl Freund, senior analyst, HPC and machine learning, Moor Insights & Strategy. “Today’s Xilinx announcements represent a significant leap forward for network architects who continue to be challenged to improve bandwidth performance of their optical networks.”

Anticipating even more demand for speed and throughput, Xilinx is demonstrating full-duplex 112G PAM4 signaling on a single lane. Industry experts regard 112Gb/s transceiver performance as necessary to address next-generation optical networking and line card densities. Customers can expect programmable devices with 112G transceivers in Xilinx’s upcoming 7nm portfolio.

Built on the most sought-after Virtex UltraScale+ class of devices for high-end applications, the new Xilinx transceiver architecture enables customers to effectively double the bandwidth capabilities of existing systems by combining the flexibility of programmable logic with 58G PAM4 transceivers. These devices can operate over existing 25G backplanes, extending the life and bandwidth of current systems while paving the way for the next generation. To help with migration, the new devices with 58G transceivers are footprint compatible with existing Virtex UltraScale+ devices in production today.

Targeted at cloud computing, 5G networking, core networks (OTN, Ethernet) and network functions virtualization (NFV) applications, this latest transceiver architecture will enable vendors to scale 50G, 100G and 400G ports and terabit interfaces in compact and less complex system designs.

Networks are changing dramatically to enable faster, more flexible and more adaptable systems, and the industry is gearing up in anticipation of the change with new optics and standards,” said Farhad Shafai, vice president, communications markets, Xilinx. “We are proud to be leading the way in delivering the most flexible and adaptable solutions to our customers, along with production-proven silicon, a track record of quality and in tandem with the ecosystem of optics, backplanes and other key technologies in development.”

Sign up for our insideHPC Newsletter

Comments

  1. Along with tranceiver bandwidth increase I hope for more advanced DSP48 technology with increased clock and bit width as well.
    Additionally it would be beneficial to have quad ported integrated ram as working with HLS shows often benefits to have more ports (than two).
    When do we get the GHz clock with FPGAs??

Leave a Comment

*

Resource Links: