Intel launched its newest third generation Ice Lake Xeon Scalable processors on Tuesday. They provide 46% increased performance and built-in AI for multiple applications including hybrid cloud, high performance computing, networking and intelligent edge.
The company claimed it already has shipped 200,000 of the processors in the first quarter of 2021. Dozens of networking equipment makers and partners are already setting up or even deploying gear that make use of the chips, Intel officials said.
In one example, Cisco announced it is using new Xeon third generation chips for three new UCS servers to deliver better performance, enhanced security and more efficiency. Cisco and Intel have worked together for more than 12 years on servers.
The new chips “could not come at a more crucial time,” said DD Dasgupta, vice president of product management for Cisco’s cloud and compute unit. “As customers’ hybrid cloud journeys accelerate, the need for simple yet powerful solutions increase.”
Intel previously launched the first third generation Xeon chip, Cooper Lake, on August 6, 2019. Cooper Lake succeeded second-generation Cascade Lake, which debuted April 2, 2019. First-generation Sky Lake chips launched July 11, 2017.
Ice Lake chips are based on 10 nm process technology with 40 cores per processor. They have the capability to provide up to 2.6 times higher performance gain even with a five-year-old system. They support up to six terabytes of memory per socket, and up to eight channels of DDR4-3200 memory per socket.
Intel tabulated 74% faster AI performance with the new chips over the prior generation. The company boasted up to 1.5 times higher performance across 20 well-known AI workloads when compared to AMD EPYC7763 and up to 1.3 times higher performance versus the Nvidia A100 GPU.
Software developers can optimize their apps using onAPI open programming. Intel makes oneAPI Toolkits.
Third-gen Xeon is designed for multiple users, including cloud. All the largest cloud service providers plan to offer services powered by the chips, Intel said.
A network-optimized N-SKU from Intel with Xeon chips will boost average performance by 62% for 5G workloads, Intel said. Also, the latest Xeon delivers 1.5 times more AI inference performance for image classification over prior generations at the intelligent edge.
Dan Rodriguez, general manager of Intel’s network platforms group, said Xeon processors are the “workhorse for our business.” He put the total silicon market potential at $25 billion by 2024.
In 2020, Intel saw an improvement in sales of 20% over 2019 from Xeon, reaching $6 billion out of Intel’s total revenue of $77.9 billion.
Networking customers want the ability to deploy 5G standalone core networks with architecture built from the ground up, Rodriguez said. They also want 5G with AI built out at the edge.
“They want to support a wide variety of workloads, both network and edge, with compelling performance,” he said in an interview.
With Xeon, Intel has shown communication service providers they can increase 5G performance by 42%. The company is working with Rakuten Mobile to use Xeon for its next-gen mobile cloud platform service. SK Telecom is pairing Xeon with Ethernet adapters and optimized NFV.
Intel Xeon is also being used in VRAN work with Verizon and Vodafone to double massive MIMO throughput, Intel said. And AT&T is using Xeon with Optane Persistent Memory to expand CDN capacity with fewer nodes for cost savings.