Nvidia Aerial

Nvidia Aerial is a software development package that allows providers to create high-performance software-defined 5G wireless radio networks with two important parameters. First, it is a low latency data link from Mellanox network interface cards to GPU memory, and second, it is a 5G physical layer signal processing engine that allows all data to be stored in high performance GPU memory.

2021: Nvidia Aerial A100 AI-on-5G

On June 28, 2021, NVIDIA announced the expansion of Arm-based CPU support for the NVIDIA Aerial A100 AI-on-5G platform, providing a wider choice of solutions in the 5G ecosystem.

According to the company, this step will help companies around the world deploy intelligent services on the periphery, and OEMs to offer standard servers with CPU based on Arm and NVIDIA AI Enterprise software with Aerial 5G.

These certified NVIDIA systems will make it more affordable to build and deploy a stand-alone vRAN that combines AI and 5G capabilities for private businesses, network hardware manufacturers, software developers, and telecommunications service providers.

The NVIDIA Aerial A100 AI-on-5G computing platform uses the NVIDIA Aerial software development kit and will include 16 Arm Cortex-A78 processors in the NVIDIA BlueField-3 A100. This standalone converged card will provide access to enterprise AI peripheral applications via 5G vRAN with Watt optimized performance and shorter deployment times.

Providing the economy of the Internet of Things BlueField-3 A100 includes an extensive library of NVIDIA software for AI and the Aerial 5G SDK, which will accelerate deployment and contribute to the development of many enterprise AI projects with low delays, including robots, automated controlled vehicles and digital twins for product development and production.

This convergence of peripheral AI and 5G technology necessitates even more diverse computations. Bluefield-3 extends the Arm ecosystem by combining optimized NVIDIA computing and Arm performance to give network providers more choice in building and deploying 5G systems. told Chris Bergey, senior vice president and director of infrastructure solutions at Arm

NVIDIA AI Enterprise provides compatibility with a specific set of accelerated CUDA applications, AI frameworks, pre-trained models, and software development kits running in a hybrid cloud. Optimization scales workloads across multiple nodes to support large deep learning models with full GPU virtualization.

In April 2021, NVIDIA announced a collaboration with Fujitsu, Google Cloud, Mavenir, Radisys and Wind River to develop solutions for the NVIDIA AI-on-5G platform.

2019: Software Presentation

On October 22, 2019, NVIDIA introduced Aerial software to accelerate 5G. Telecommunications companies can use the NVIDIA Aerial software development suite and ready-to-use servers with NVIDIA GPU to create programmable and scalable 5G software-defined networks to enable AI services on end devices.

Founder and CEO of NVIDIA Jensen Huang said that Aerial software, based on the NVIDIA EGX platform, allows you to create AI services and immersive content on peripherals of 5G networks.

As noted in NVIDIA, 5G offers high speeds, reducing delays by 10 times and increasing bandwidth by 1000 times, which will cause the emergence of millions of connected devices per square kilometer. 5G also gave rise to an important concept of network segmentation, which will allow the telecom to dynamically - from session to session - offer unique services to customers. Traditional solutions could not be reconfigured quickly, so the telecom needs a network architecture with high performance and the ability to reconfigure on the fly.

Such virtualized radio access networks operate in a wireless infrastructure close to customers, and are therefore suitable for providing AI services on end devices. They are an important element for creating a modern 5G infrastructure that can run different applications that dynamically appear on a common platform, NVIDIA emphasized.

According to the developer, thanks to NVIDIA Aerial, the same computing infrastructure required for 5G networks can be used to provide AI services, such as smart smart cities manufacturing ,/and ARVR. cloudy geyming

Aerial includes two important SDKs - CUDA Virtual Network Function (cuVNF) and CUDA Baseband (cuBB) to simplify the creation of highly scalable and programmable 5G software-defined radio networks using ready-to-use NVIDIA GPU-based servers.

• The NVIDIA cuVNF SDK provides optimal I/O and batch processing by sending 5G packets directly to GPU memory from GPUDirect-enabled network interface cards.
• The NVIDIA cuBB SDK provides a GPU-accelerated 5G signal processing pipeline, including cuPHY for L1 5G Phy, which provides high transfer speed and efficiency by performing physical layer processing in high-performance GPU memory.

The NVIDIA Aerial SDK is based on the NVIDIA EGX stack, providing GPU acceleration to the operator-class Kubernetes infrastructure. The NVIDIA EGX stack includes the NVIDIA driver, the NVIDIA Kubernetes plugin, the NVIDIA container runtime environment, and the NVIDIA GPU monitoring software.

To simplify the management of GPU servers, telecom companies can install all the necessary NVIDIA software as containers running on Kubernetes, open source software that is widely used to accelerate the deployment and management of complex software of any kind.

According to the developer, Aerial provides high investment efficiency due to the high adaptation to changes in network traffic during the day, as well as the ability to provide services that meet the needs of users.

As of October 2019, NVIDIA Aerial is already available to partners. Mass availability is scheduled for the end of 2019.