Intel has officially launched its 4th Gen Intel Xeon Scalable platform.
This is the official production release for the SoC and software platform known as Sapphire Rapids that has been previously trailed and outlined by Intel. In the telco space, the launch is important for Intel because telcos have doubted that x86 hardware is suitable for the most intensive vRAN processes, such as massive MIMO. As some operators and vendors look to enagage with ARM-based SoCs that offer inline Layer 1 acceleration, Intel sees 4th Gen Xeon as its tool to fight back and prove that there is an available Intel platform that can meet operator requirements around power efficiency and performance.
The company says that it has been redesigned with a “workload first” architecture that will allow developers and manufacturers to meet specific use cases that have intensive requirements for processing, power efficiency and sustainability. The platform comes with 12 built-in accelerator engines for deep learning, analytics, load balancing, encryption and data streaming.
One dedicated accelerator, Intel Advanced Vector Extensions 512 (Intel AVX-512) for vRAN is designed to deliver greater capacity for vRAN workloads. It claims to have doubled vRAN performance compared to its previous generation platform, within the same power envelope.
This is the platform that TMN revealed was being adopted by Rakuten Mobile in the next phase of its cloud-native network rollout. Today’s release was backed by official endorsement from Telefonica, which said it was going to use the platform to support its “most advanced” Open RAN and CNF (Cloud Native Function) rollouts later this year.
A written statement from Telefonica said, “Telefonica looks forward to the improved performance and features with 4th Gen Intel Xeon Scalable processors when it deploys in its network later this year.”
CTO Enrique Blanco, in conversation with Intel’s Dan Rodriguez, GM Network and Edge Solutions, said that 4th Gen Xeon is the “magic weapon helping us to meet our final architecture.”
Blanco said, “When I see Open RAN, vRAN, CNF – reducing power and energy consumption and going through additional throughput, I can only get this if I am using Intel’s 4th generation.
“It is providing me the technology I need: how can I fully virtualise, open all the interfaces and connect hundreds of containers to a centralised approach? This for me is the fourth generation of Xeon.”
Blanco added, “This long term partnership means trust. In our most advanced step in Open RAN capabilities, we trust in Intel.”
Ericsson sees this as a big step forward in the industry’s transition to a cloud-based paradigm
Intel said today that a broad range of companies had already committed, including hardware providers Dell and HPE, hyperscalers and cloud providers as well as telecoms equipment providers.
Ericsson, which has been increasingly talking up its vRAN capabilities over the past 12 months, is another that endorsed the launch. A statement said, “Ericsson sees this as a big step forward in the industry’s transition to a cloud-based paradigm. Intel’s optimised performance and power efficiency, combined with the scalability and flexibility of Ericsson Cloud RAN, helps enable high-capacity solutions for RAN environments. Ericsson looks forward to seeing deployments broadly within its customers’ networks.” In December 2022, Ericsson authored a joint paper with Verizon stating that their preferred solution for vRAN was lookaside acceleration.
Intel said that it has increased its core-per-socket count, added built in accelerators and optimised software to improve performance and sustainability. It has previously claimed that built-in acceleration is akin to the inline acceleration, in that it doesn’t involve the data-heavy interface between CPU and accelerator necessitated by its previous “lookaside” accelerator architecture.
4th Gen Intel Xeon Scalable processors feature a new architecture with higher per-core performance than the previous generation. They also feature up to 60 cores per socket and one, two, four, or eight sockets per system. To balance those core-count increases, the platform provides accompanying advances in the memory and input/output (I/O) subsystems. DDR5 memory provides up to 1.5x the bandwidth and speed of DDR4, for 4,800 megatransfers per second (MT/s). The platform also features 80 lanes of PCIe Gen5 per socket, for improved I/O compared to earlier platforms.
The company said that built-in acceleration is an alternative, more efficient way to achieve higher performance than growing the CPU core count.
A blog post from Rodriguez, who heads up Intel’s network and edge solutions business, said, “4th Gen Intel Xeon Scalable processors deliver our most powerful compute for demanding edge and network workloads. It provides significant improvements in per-core performance, introduces DDR5 memory, PCIe 5.0 and increased I/O support. And they offer the most built-in accelerators of any CPUs on the market.
“Network operators get a solution to deliver sophisticated packet processing with energy-efficient performance to power virtualized network functions anywhere. They also get an agile platform to handle dramatic 5G traffic growth from user-generated content and new media types such as AR/VR.
“By virtualizing their 5G networks end-to-end and standardizing on Intel architecture, network providers get a converged service advantage, running all core, edge and RAN workloads flexibly on a common platform. For 5G core workloads, built-in accelerators help increase throughput and decrease latency, while advances in power management enhance both the responsiveness and the efficiency of the platform. And with 4th Gen Intel Xeon Scalable processors, we are delivering about twice the vRAN capacity compared to the prior generation, without increasing power consumption.”