- Arm Neoverse V2, code-named Demeter, a new processor, is anticipated to play a significant role in data centers for years to come.
- Compared to servers powered by conventional architectures, Grace is supposed to give twice the performance per watt by combining the power efficiency of V2 with LPDDR5X memory.
Arm Ltd., a British semiconductor company, revealed an updated product roadmap featuring new processor designs that, according to the company, will serve as a foundation for various workloads over the following years.
Arm’s roadmap sees the addition of Arm Neoverse V2, code-named Demeter, a new processor that is anticipated to play a significant role in data centers for years to come. The business said its most recent core is designed to offer the best per-thread performance for cloud, hyperscale, and high-performance computing workloads.
In an online presentation, Rene Haas, the Chief Executive of Arm, stated that Arm Neoverse had been chosen as the compute foundation for numerous next-generation infrastructure platforms by industry leaders in the cloud, HPC, 5G, and edge computing. To compete with Intel Corp.’s Xeon processors and Advanced Micro Devices Inc.’s Epyc chips, the company introduced Arm Neoverse in 2018. Since then, it has developed into three distinct platforms, the Arm V-series, N-series, and E-series, each of which is targeted at different compute tasks.
Dermot O’Driscoll, Vice President of Product Solutions at Arm, stated at the presentation that more than 20 chipmakers had embraced the V-series chip designs. According to him, the blueprints give chipmakers the ability to quickly develop new solutions and respond to the market’s and customers’ changing needs.
O’Driscoll added that the future cloud infrastructure must accommodate a surge in new data while efficiently handling applications that are getting more and more complicated. On the other hand, customers also want to see higher power efficiency to reduce their carbon footprints.
According to him, customers seek solutions that increase cloud workload performance without requiring more space or power. He said that the company’s response to these needs is the Arm’s Neoverse V2 platform, which would provide industry-leading integer performance for cloud and HPC applications.
Additionally, it will be arriving soon. According to Arm, numerous partners are working on getting their chip designs based on Neoverse V2 to the market. One is Nvidia Corp., which built its new Grace data center central processing unit on the V2 platform. Compared with servers powered by conventional architectures, Grace is supposed to give twice the performance per watt by combining the power efficiency of V2 with LPDDR5X memory.
According to Arm, Nvidia’s Grace CPUs are now being produced, and many other clients are also working on devices based on the V2 architecture. Ampere Computing LLC, Marvell Technologies Inc., and Amazon Web Services Inc. are a few clients.
Patrick Moorhead, Analyst of Moor Insights & Strategy, said, “The new datacenter roadmap is the most aggressive I have ever seen from Arm. Arm is projecting it will have the highest-performance CPU versus AMD’s or Intel’s equivalent CPUs. As I have said, industry-standard architecture does not connote performance but rather a design. Arm reinforced that today.”
There isn’t much information about the upcoming N-series and E-series chip designs, which are scheduled for launch in 2023. The N-series chips are made for applications where thread count is valued over single-threaded performance, but the V-series cores are built to push performance limits. The Arm E-series cores, on the other hand, are designed for acceleration and data plane processing applications like edge networking and 5G RAN.
According to Arm, the third version of the N-series core designs is now under development and will be available to partners later in 2023. The E2’s replacement is currently being created and will also be released in 2023.