Cadence enhances cloud datacenter infrastructure with its initial silicon-proven, long-reach 7nm 112G SerDes IP

Cadence Design Systems unveiled on Monday its initial silicon-proven, long-reach 112G SerDes IP in 7nm. The Cadence 7nm 112G PAM-4 SerDes IP delivers power, performance and area (PPA) efficiency required to build high-port density networking products for next-generation cloud-scale and telco datacenters.

Cadence is now ready to engage broadly with customers to enable their next-generation high-performance computing (HPC) ASICs, machine learning accelerators, and switch fabric SoCs.

112G SerDes technology doubles the data rate of 56G SerDes, meeting the exploding high-speed connectivity needs for emerging data-intensive applications such as machine learning and neural networks. However, the long-reach connections needed in advanced server and networking equipment are notoriously difficult to design.

Channel loss for long-reach applications at the 112G Nyquist frequency is much greater than medium-reach applications, demanding novel approaches to SerDes design.

Based on technology gained from Cadence’s 2017 acquisition of nusemi, the long-reach 112G SerDes supports backplane, copper and optical connections. Key benefits include unique firmware-controlled adaptive power design provides optimal power and performance tradeoffs and more efficient system designs based on platform requirements; DSP-based architecture provides superior data recovery for lossy and noisy channels; with extended reach capability enables customers to use lower cost PCBs and achieve greater flexibility in PCB and system design.

It also delivers multi-rate support, including 112/56Gbps PAM-4 (four-level pulse amplitude modulation) and 56/28/10Gbps NRZ (non-return-to-zero) data rate, provides backward compatibility with legacy equipment operating at lower speeds, and fully autonomous startup and adaptation, as well as an integrated BIST (built-in self-test) capable of producing and checking PRBS (pseudo random binary sequence), are supported to enhance IP ease of use.

Escalating mobile data consumption, burgeoning AI and machine learning applications, and emerging 5G communications requirements demand ever-increasing bandwidth, straining the existing cloud datacenter server, storage and networking infrastructure.

Early adopters in the high-end cloud datacenter market are now installing 400G Ethernet ports, with 400G Ethernet expected to go mainstream in 2020 as early adopters begin 800G Ethernet deployment. 112G SerDes technology doubles the data rate of 56G SerDes, meeting the exploding high-speed connectivity needs for emerging data-intensive applications such as machine learning and neural networks.

“The 112G SerDes is a new and critical enabling technology that allows the industry to build out the next-generation 100G, 400G and 800G Ethernet cloud infrastructure more rapidly and cost effectively,” said Lip-Bu Tan, chief executive officer of Cadence. “Our silicon-proven 112G long-reach multi-rate SerDes IP places Cadence at the forefront of high-performance computing system design enablement. By enabling 100Gb/sec per lane, Cadence’s solution reduces the lane count, heat dissipation and cost required to build the next generation of hyperscale infrastructure.”

Early adopter customers have already begun to design the new SerDes into their SoCs, and Cadence is available for broad customer engagements now.

 


IoT Innovator Newsletter

Get the latest updates and industry news in your inbox! Enter your email address and name below to be the first to know.

Name