More AI for cell networks as Intel rolls out vRAN platform

MWC Intel’s opening day at MWC is all about networks and edge compute, specifically a dedicated edge platform, a developer kit for AI models for vRAN, and a power management tool for 5G infrastructure.

Intel's Edge Platform, which the Santa Clara giant indicates will be generally available this quarter, was created to help organizations deploy and manage edge applications.

Despite the announcement, Intel didn't give much away about the platform, other than it uses "the Intel ecosystem," meaning it runs on Intel hardware, and that it was designed to offer cloud-like scalability and run on "brownfield" or "greenfield" hardware – meaning customers necessarily have to lash out for new kit to use it.

This is an evolution of Intel's Project Strata, introduced at the company's Innovation conference last year, according to Intel VP and GM of Network and Edge Software Pallavi Mahajan.

Naturally, it also has built-in AI model development, management and inferencing support, she claimed. It will be sold through Intel's channel, which will provide the customer support, and pricing will depend upon the capabilities customers make use of.

Intel also announced its Infrastructure Power Manager at last year's MWC, and this week declared the product generally available.

According to the chipmaker, it uses the built-in telemetry of Xeon processors to match server power consumption to the volume of network data traffic without compromising on performance indicators such as throughput, latency or packet drop. It can give power savings of up to 30 percent, Intel claimed.

"We're committed to helping our partners and customers improve overall power and energy savings to reach their sustainability goals,' said Intel's VP and GM of the Network and Edge Solutions Group Dan Rodriguez.

Also announced is early availability of a vRAN AI Developer Kit, to help telecoms operators incorporate AI features into their network.

Operators can run AI workloads on existing general purpose servers using Xeon chips, according to Rodriguez, and the vRAN AI Developer Kit allows the coders to build, train and deploy them in a vRAN environment. The platform is built on top of "Intel AI-optimized software libraries, frameworks and tools."

But why would network operators want to do this? Use cases include optimizing performance and using machine learning to manage resources. Intel said it is working with operators on proof-of-concept RAN AI use cases, and will demonstrate the tech at MWC.

• Cloud, datacenter vendors muscle in on traditional telco territory at MWC
• Nope, still no Intel Sapphire Rapids Xeon processors for you right now
• Years late and 36 cores short of AMD, who are Intel’s 4th-gen Xeons even for?
• Intel targets edgy 5G and AI with new chips, software

Intel is also previewing the Sierra Forest Xeon chip at MWC, which is due to be available sometime later this year. The chip, which features up to 288 of Intel's energy efficient E cores, is described as being "well-suited for 5G core workloads."

Granite Rapids-D, a version of the Granite Rapids server chips due later this year aimed at edge and telco infrastructure like other Xeon-D parts, was also announced ar MWC. Although Granite Rapids-D isn't shipping until 2025, Intel said it is sampling now to system builders. ®