• The company wants to modernize its customers’ 5G, edge, and enterprise infrastructures by bringing “AI Everywhere.”
  • According to Intel, Granite Rapids-D will significantly increase those performance advantages, allowing customers to lower their vRAN expenses and deploy such networks on a far larger, worldwide scale.

Less than a week after announcing intentions to improve the manufacturing of next-generation computer chips, Intel Corp. has introduced two new silicon platforms built on top of its most potent current chip architecture.

The new processors include Sierra Forest, which will be released in the second half this year, and Intel’s new Xeon processor, code-named Granite Rapids-D, which will be available in 2025. Recently, they revealed the MWC 2024 along with a new Edge Platform intended to handle workloads related to artificial intelligence (AI), applications, and infrastructure deployed at the network’s edge.

According to Intel, the new central processing units will help clients take advantage of the  “emerging opportunities” brought about by automation and AI, thereby lowering the total cost of ownership and boosting the operational effectiveness of a wide range of next-generation applications. Ultimately, the company wants to modernize its customers’ 5G, edge, and enterprise infrastructures by bringing “AI Everywhere.”

Granite Rapids-D

Building on the success of its 4th Gen Intel Xeon Processors with Intel vRAN Boost, also known as Sapphire Rapids EE, the upcoming Xeon processor, Granite Rapids-D, will help Intel maintain its leading position in virtual radio access network workloads.

According to Intel, the older chips reduced vRAN compute power usage by up to 20% while doubling the capacity of vRAN workloads compared to its previous generation silicon. This allowed network operators to support or run twice as many cell sites or customers.

Granite Rapids-D will significantly increase those performance advantages, allowing customers to lower their vRAN expenses and deploy such networks on a far larger, worldwide scale. Granite Rapids-D will reportedly use integrated Intel vRAN Boost acceleration and Intel AVX for vRAN to attain those benefits.

Laboratory samples of the Granite Rapids-D chips are presently being collected by Samsung Electronics Co. Ltd. and Ericsson AB, among others. Furthermore, Intel is collaborating with various partners, including Red Hat Inc., Dell Technologies Inc., Hewlett-Packard Enterprise Co., and Lenovo Group Ltd., to guarantee market preparedness prior to the launch of its product the following year.

Intel is granting early access to its vRAN AI Development Kit to select partners prior to the launch of Granite Rapids-D. It is stated that the vRAN AI Development Kit assists clients in constructing, training, optimizing, and deploying general-purpose server-capable AI models for vRAN use cases.

Customers should be able to accommodate new AI workloads by utilizing their current vRAN network architecture. Together with Intel’s 4th generation Intel Xeon processors, the optimized AI models in the vRAN AI Development Kit enable users to create AI-powered networks that can dynamically reconfigure to save costs and accommodate new revenue streams, according to the company.

Sierra Forest

While customers may have to endure a temporary delay in obtaining the Granite Rapids-D chips, the Sierra Forest processors will be available much sooner, Intel announced recently. The Sierra Forest processors are scheduled to debut in the latter part of the year.

The company promised that the Sierra Forest chips, which will include up to 288 cores and employ its most recent E-core technology, would enable vRAN network operators to attain a 2.7-times increase in performance per rack for 5G workloads.

Furthermore, the Sierra Forest CPUs will be included with an upgraded version of Intel Infrastructure Power Manager software, which made its debut last year. This software enables users to use integrated telemetry to lower power consumption without experiencing a decline in essential performance metrics like throughput, latency, and packet drop.

According to Intel, the new processors and Power Manager software have piqued the interest of network operators and ecosystem partners, including BT Group, Dell, Ericsson, HPE, Lenovo, KDDI Corp., and SK Telecom Co., Ltd.

Intel Edge Platform

AI at the network edge presents customers with new opportunities, according to Intel, which stated it is utilizing its extensive installed base of over 90,000 edge deployments and 200 million processors distributed over the past decade to assist customers in capitalizing on these prospects.

The organization elaborated on its new Edge Platform, which will empower clients to profit from the enormous volumes of data produced at the network edge. This knowledge can then be utilized to improve customer experiences and automate the expansion of business operations.

According to the company’s proposal, Intel Edge Platform is an open-source, modular platform that will simplify the development, deployment, operation, and management of edge and AI applications for businesses in a manner akin to the cloud. It is said to accommodate heterogeneous components, which can lead to a cheaper total cost of ownership (TCO) and allow policy-based, zero-touch control of edge apps and infrastructure and AI workload at the edge. To provide real-time AI inferencing optimization and dynamic workload distribution throughout the customers’ infrastructure, it will also offer AI runtime with OpenVINO inference.

As per Intel’s statement, the Edge Platform will enable developers to fine-tune application orchestration, enabling them to remotely allocate latency-sensitive workloads to the most optimal device in order to guarantee optimal application performance. It will also facilitate the development of AI models and applications with hybrid capabilities from the edge to the cloud, utilizing both low-code and high-code approaches. In turn, this will enable developers to construct vertical, industry-specific services for applications such as video, time-series data, and digital twins, as well as horizontal edge services such as data annotation for creating new AI models.

Before the end of the current quarter, Intel stated that the Edge Platform will roll into general availability.