Less than a week after Intel Corp. unveiled plans to advance its manufacturing process for next generation computer chips, it has announced two new silicon platforms based on its most powerful existing chip architecture.

The new chips include Intel’s latest Xeon processor, code-named Granite Rapids-D, which will become available in 2025, plus Sierra Forest, which will launch in the second half of this year. They were announced a MWC 2024 early Monday local time alongside a new Edge Platform that’s designed to manage infrastructure, applications and artificial intelligence workloads deployed at the edge of the network.

Intel said the new central processing units will enable customers to capitalize on the “emerging opportunities” enabled by AI and automation, helping to improve the total cost of ownership and operational efficiency of a host of next-generation applications. Ultimately, the company aims to bring “AI Everywhere” in order to modernize its customers’ 5G, edge and enterprise infrastructures.

Granite Rapids-D

The future Xeon processor, Granite Rapids-D, will help Intel to extend its leading position in virtual radio access network workloads, building on the success of its 4th Gen Intel Xeon Processors with Intel vRAN Boost, otherwise known as Sapphire Rapids EE.

Intel explained that those older chips doubled the capacity of vRAN workloads compared with its prior generation silicon, enabling network operators to double the number of cell sites they operate or subscribers they can support, while reducing vRAN compute power consumption by up to 20%.

Granite Rapids-D will add to those performance gains significantly, Intel said, enabling customers to drive down their vRAN costs and deliver those networks on a much wider, global scale. According to Intel, Granite Rapids-D will leverage Intel AVX for vRAN and integrated Intel vRAN Boost acceleration to achieve those gains.

The Granite Rapids-D chips are currently being sampled in labs by customers including Samsung Electronics Co. Ltd. and Ericsson AB. In addition, Intel is also working with partners such as Dell Technologies Inc., Hewlett-Packard Enterprise Co., Lenovo Group Ltd., Red Hat Inc. and others to ensure market readiness in time for next year’s launch.

Ahead of Granite Rapids-D’s debut, Intel is releasing its vRAN AI Development Kit in early access to select partners. The vRAN AI Development Kit is said to help those customers build, train, optimize and deploy AI models for vRAN use cases that can be run on general purpose servers.

The idea is that customers will be able to leverage their existing vRAN network architecture to support new AI workloads. The optimized AI models within the vRAN AI Development Kit can be combined with Intel’s 4th Gen Intel Xeon processors to help customers build AI-powered networks that can reconfigure themselves dynamically to optimize costs and support new revenue streams, the company said.

Sierra Forest

Although customers will have to wait a while to get their hands on the Granite Rapids-D chips, they’ll be able to access the Sierra Forest processors much sooner, as they’re slated to launch in the first half of the year, Intel said today.

The Sierra Forest chips will feature up to 288 cores and utilize the company’s latest E-core technology to help vRAN network operators to achieve a 2.7-times performance boost per rack for 5G workloads, the company promised.

In addition, the Sierra Forest CPUs will also feature an updated version of the Intel Infrastructure Power Manager software that first debuted last year, which allows customers to take advantage of built-in telemetry to reduce power consumption without suffering any drop in key performance metrics such as throughput, latency and packet drop.

Intel said network operators and ecosystem partners including BT Group, Dell, Ericsson, HPE, Lenovo, KDDI Corp. and SK Telecom Co., Ltd. have all shown a key interest in the new chips and Power Manager software.

Intel Edge Platform

Last but not least, Intel said it’s leveraging its expansive installed base of more than 90,000 edge deployments and 200 million processors sold over the last 10 years to help customers take advantage of new opportunities that lay in deploying AI at the network edge.

The company detailed its new Edge Platform, which will enable customers to leverage the massive amounts of data generated at the network edge, and use this information to enhance customer experiences and scale their business operations via automation.

Intel Edge Platform is a modular, open-software platform that will make it simpler for enterprises to develop, deploy, run and manage edge and AI applications with cloud-like simplicity, the company said in its pitch. It’s said to support heterogeneous components that can result in a lower TCO and enable zero-touch, policy-based management of edge infrastructure and applications, plus AI workload at the edge. It will also support AI runtime with OpenVINO inference to enable real-time AI inferencing optimization and dynamic workload placement across the customers’ infrastructure.

According to Intel, developers will be able to use the Edge Platform to fine-tune application orchestration so they can remotely place latency-sensitive workloads on the most optimal device to ensure the best application performance. It will also enable both low-code and high-code AI model and application development with hybrid capabilities from the edge to the cloud. In turn, this will help developers to create horizontal edge services such as data annotation for building new AI models, as well as vertical, industry-specific services for applications such as video, time-series data and digital twins.

Intel said the Edge Platform will launch in general availability before the end of this quarter.

Images: Intel