Arm Flexes Next-Gen Neoverse N3 And V3 With Huge AI Performance Uplifts
Arm’s latest Neoverse chip technology is based on the Arm computing architecture, and this announcement simply ushers in this new Arm IP as now available for companies to license to create chips for the consumer, data center and enterprise markets. The company also provided a couple of codenames for next-gen CSS V-series (Vega) and N-series (Ranger) platforms, which some believe will eventually become the Neoverse V4 “Adonis” and N4 “Dionysus” products.
According to Arm, its Neoverse CSS V3 technology offers the “fastest path from concept to deployment of next generation cloud compute for its global partners.” It also offers a “significant performance increase," better efficiency, and differentiated features.
The Arm Neoverse CSS V3 product will be for the high-performance V-series. Arm touts it as being a 50% improvement in performance-per-socket over the last generation CSS N2. The Neoverse CSS V3 is 64 cores per cluster, with up to 128 cores per socket, while supporting memory and IO features such as PCIe Gen5, CXL 3.0, and HBM3.
The company also proclaimed that it delivers “the undifferentiated heavy lifting, enabling partners to build specialized silicon at a lower cost, with less risk and a faster time-to-market compared to traditional development methods.” It added, “At the heart of this CSS is the Neoverse V3 CPU.”
Perhaps the most interesting capability of this new architecture is that the company says this technology can help provide a NVIDIA Grace Hopper style compute platform, for bespoke solutions, if customers wish to attach their own AI accelerator.
Another aspect of the announcement surrounds the Neoverse CSS N3, as part of the N-series CSS roadmap. This Arm technology is said to provide a 20% increase in performance-per-watt compared to CSS N2. The platform will now offer up to 32 cores and as little as 8 cores, while the 32-core variant can operate down to as little as a 40W TDP.
Dermot O’Driscoll, VP of Product Solutions, Infrastructure Line of Business, Arm, remarked, “We tune CSS N3 to meet a gap we see in the market for highly efficient compute that meets infrastructure performance requirements.”
While Arm’s announcement is exciting, it is not expected to be just the announcement that will drive companies en masse toward the Arm architecture. However, Mohamed Awad SVP/GM, Infrastructure Business at Arm explained, “The platforms we have introduced today are the best generation of Arm Neoverse yet. They’ll be the foundation of the next-generation products and services partners build as more AI workloads are adopted and permeate across the entire continuum of compute.” Awad’s confidence in the Arm ecosystem can be felt in his remark that he firmly believes the future of compute, including AI, “will be built on Arm.”
Shining even more light on the Arm architecture today, was an announcement from Intel that it would be "supporting startups in developing Arm-based technology and offering essential IP, manufacturing support and financial assistance to foster innovation and growth." The company also added that it will be collaborating with Arm to "provide cutting-edge foundry services for Arm-based system-on-chips (SoC)."
To read more on Arm’s announcement, you can visit the company’s website here.
