NVIDIA Unveils GPU-Accelerated AI-on-5G System for Edge AI, 5G and Omniverse Digital Twins

Telcos are searching for industry-standard options that may run 5G, AI functions and immersive graphics workloads on the identical server — together with for laptop imaginative and prescient and the metaverse.

To satisfy this want, NVIDIA is creating a brand new AI-on-5G answer that mixes 5G vRAN, edge AI and digital twin workloads on an all-in-one, hyperconverged and GPU-accelerated system.

The decrease value of possession enabled by such a system would assist telcos drive income progress in good cities, in addition to the retail, leisure and manufacturing industries, to help a multitrillion-dollar, 5G-enabled ecosystem.

The AI-on-5G system consists of:

• Fujitsu’s virtualized 5G Open RAN product suite, which was developed as a part of the 5G Open RAN ecosystem expertise (OREX) challenge promoted by NTT DOCOMO. It additionally consists of Fujitsu’s virtualized central unit (vCU) and distributed unit (vDU), plus different virtualized software program capabilities of vRAN from Fujitsu.
• The NVIDIA Aerial software program growth equipment for 5G vRAN; NVIDIA Omniverse for constructing and working customized 3D pipelines and large-scale simulations; NVIDIA RTX Digital Workstation (vWS) software program; and NVIDIA CloudXR for streaming prolonged actuality.
• {Hardware} consists of the NVIDIA A100X and L40 converged accelerators.

OREX has supported efficiency verification and analysis checks for this method.

Collaborating With Fujitsu

“Fujitsu is delivering a completely virtualized 5G vRAN along with multi-access edge computing on the identical high-performance, energy-efficient, versatile and scalable computing infrastructure,” mentioned Masaki Taniguchi, senior vice chairman and head of cellular programs at Fujitsu. “This mix, powered by AI and XR functions, allows telcos to ship ultra-low latency providers, extremely optimized TCO and energy-efficient efficiency.”

The announcement is a step towards conducting the O-RAN alliance’s objective of enabling software-defined, AI-driven, cloud-native, totally programmable, energy-efficient and commercially prepared telco-grade 5G Open RAN options. It’s additionally in step with OREX’s objective of implementing a extensively adopted, high-performance and multi-vendor 5G vRAN for each public and enterprise 5G deployments.

The all-in-one system makes use of GPUs to speed up the software-defined 5G vRAN, in addition to the sting AI and graphics functions, with out bespoke {hardware} accelerators nor a selected telecom CPU. This ensures that the GPUs can speed up the vRAN (primarily based on NVIDIA Aerial), AI video analytics (primarily based on NVIDIA Metropolis), streaming immersive prolonged actuality (XR) experiences (primarily based on NVIDIA CloudXR) and digital twins (primarily based on NVIDIA Omniverse).

“Telcos and their clients are exploring new methods to spice up productiveness, effectivity and creativity by way of immersive experiences delivered over 5G networks,” mentioned Ronnie Vasishta, senior vice chairman of telecom at NVIDIA. “At Cellular World Congress, we’re bringing these visions into actuality, showcasing how a single GPU-enabled server can help workloads corresponding to NVIDIA Aerial for 5G, CloudXR for streaming digital actuality and Omniverse for digital twins.”

The AI-on-5G system is a part of a rising portfolio of 5G options from NVIDIA which are driving transformation within the telecommunications {industry}. Anchored on the NVIDIA Aerial SDK and A100X converged accelerators — mixed with BlueField DPUs and a set of AI frameworks — NVIDIA gives a high-performance, software-defined, cloud-native, AI-enabled 5G for on-premises and telco operators’ RAN.

Telcos working with NVIDIA can achieve entry to hundreds of software program distributors and functions within the ecosystem, which may help handle enterprise wants in good cities, retail, manufacturing, industrial and mining.

NVIDIA and Fujitsu will exhibit the brand new AI-on-5G system at Cellular World Congress in Barcelona, working Feb. 27-March 2, at corridor 4, stand 4E20.