Intel Corp. is pushing to be at the forefront of the rollout of next-generation 5G networks, launching new hardware and software today, including its first 10-nanometer system-on-chip for wireless base stations.

The new Intel Atom P5900 processor is the company’s first SoC designed for wireless base stations, which are radio receivers and transmitters that serve as a hub for local wireless networks.

Experts say wireless base stations are a key early deployment target for 5G providers as they will be critical to enabling speedier wireless networks. That’s why Intel is going all-out to dominate the market with the Atom P5900, which is designed to to provide both higher bandwidth and lower latency than existing base station silicon.

In a media briefing, Dan Rodriguez, corporate vice president and general manager of Intel’s Network Platforms Group, said partners that include Telefonaktiebolaget LM Ericsson and ZTE Corp. will begin deploying the Atom P5900 this year. Ultimately, he said, the company intends to establish itself as the market leader in wireless base stations by 2021.

Alongside the new Atom P5900, the company also launched a new breed of Intel Xeon Scalable processors for use in 5G network deployments. One of the most notable changes with 5G networks is that far more data will be processed at the edge of the network, where lower latency translates to faster insights and more business value.

“5G is a massive inflection point, driving data and new data-driven services,” Rodriguez said.

Intel said the new 2nd Gen Intel Xeon Scalable processors are ideal for this kind of task, offering an average of 42% more performance per dollar than its previous generation chips.

“This is the fastest-ramping Xeon in history,” said Lisa Spelman, corporate vice president and general manager of Intel’s Xeon processor and data center marketing.

Spelman said the 2nd Gen Intel Xeon Scalable processor is well-suited for edge artificial intelligence workloads too, since it incorporates the company’s Deep Learning Boost technology that helps it to provide up to six times the performance of competing CPU platforms. The processors have already been deployed by customers including Amazon Web Services Inc., Microsoft Corp. and Alibaba in their public cloud platforms, she said.

Staying with AI, Intel also introduced a brand new structured application-specific integrated circuit, or ASIC, that can be customized for specific workloads. Structured ASICs are an alternative to field-programmable gate arrays, which are hardware accelerators that can be reprogrammed on the fly for different tasks. ASICs are just as customizable, but they can however only be programmed once. Still, they are much more power efficient, and more suitable for high-volume mass production.

Intel rounded out its 5G networking portfolio with the launch of its new Ethernet 700 Series Network Adapter that comes with “hardware-enhanced precision time protocol.” Rodriguez said this last feature provides the ability to maintain accurate time synchronization across networks, which is critical for eliminating latency in financial services, video streaming and emergency services.

The adapter, Intel said, boosts the timing precision required for 5G networks through a combination of hardware and software enhancements. It’s sampling now and will go into production in the second quarter.

Finally, Intel said it’s expanding its Open Network Edge Services Software toolkit. OpenNESS is an open-source reference toolkit that enables the ecosystem to create and deploy new edge applications and services. From today, it now supports standalone 5GNR and Enhanced Platform Awareness or EPA deployments, which should help customers to deploy edge-based microservices more easily.

Analyst Patrick Moorhead of Moor Insights & Strategy said today’s announcements highlight Intel’s timely efforts to diversify its business rapidly into new growth areas when carriers are in the middle of a transition to a new network architecture.

“What I like about Intel’s approach for 5G carriers is that it has nearly every kind of silicon you may need, from CPUs adndGPUs to FPGAs, soft ASICs and hardened ASICs,” Moorhead said. “Also, it’s at many different layers from the cloud to core to edge.”

That’s just as well because wireless carriers need to make massive investments in their infrastructure in order to deliver true 5G networks, added Holger Mueller of Constellation Research Inc.

“Current infrastructure cannot handle either the volume or the performance, so a massive investment is key,” Mueller said. “Intel wants a piece of that and is doing so with infrastructure, chips and software.”

With reporting from Robert Hof

Image: mohamed_hassan/Pixabay