Nvidia renews efforts in Edge AI for smart cities with new solutions and partnerships
Nvidia recently concluded its annual GTC Conference for AI during which it unveiled a series of product releases and new partnerships aimed at driving further growth in its efforts to bring edge artificial intelligence (AI) to more users. Specifically, the company released an AI supercomputer for embedded computing applications at the edge and partnered with Mavenir and Siemens on 5G solutions for smart cities. In addition, Nvidia has also welcomed Lanner to its virtual AI conference and joined forces with Elma to develop a rugged platform for AI-based applications.
Nvidia introduces Jetson AGX Orin
According to Nvidia, the new device has been designed specifically for robotics, autonomous machines, medical devices, and other embedded computing applications at the edge.
Jetson AGX Orin is built on the Nvidia Ampere architecture, and can reportedly provide more than six times the processing power over its predecessor Jetson AGX Xavier while retaining the same form factor and pin compatibility. The device can deliver 200 trillion operations per second, and features an NVIDIA Ampere architecture GPU, together with Arm Cortex-A78AE CPUs.
The new embedded computing partner ecosystem is compatible with Nvidia’s Cuda-X accelerated computing stack, as well as the company’s application development and optimization tools.
Software frameworks for Jetson AGX Orin include Nvidia Isaac Sim on Omniverse for robotics, Nvidia Clara Holoscan SDK for healthcare, and Nvidia Drive for autonomous driving.
The Nvidia Jetson AGX Orin module and developer kit are scheduled for release in the first quarter of 2022.
Mavenir announces MAVedge-AI Intelligent Video Analytics
Mavenir is offering a new product that will be powered by the Nvidia Metropolis platform and will enable firms to deploy AI and 5G together like the Nvidia Metropolis intelligent video analytics (IVA) solution using an edge server with a converged accelerator.
MAVedge-AI will utilize the Nvidia Aerial software development kit and a converged accelerator card that combines either the NVIDIA A100 or A30 GPU and the NVIDIA BlueField-2 data processing unit (DPU), alongside Nvidia’s ‘5T for 5G’ solution.
The move aims to open up new use cases for AI at the edge by delivering reliable and secure AI applications over 5G networks.
Together with the new MAVedge-AI, Mavenir also announced a new IVA solution powered by Nvidia Metropolis.
MAVedge-AI will be officially released in the first half of 2022, while the intelligent video analytics solution powered by Nvidia Metropolis is already available for purchase.
Nvidia and Mavenir are reportedly collaborating with Dell and Foxconn to facilitate the deployment of AI-on-5G edge computing infrastructures.
Siemens Energy deploys Nvidia Triton Inference Server
Siemens Energy announced it has deployed the Nvidia Triton Inference server as part of a partnership with Nvidia that will focus on using AI to facilitate the transition to autonomous power plants and a consequent reduction in costs. Triton is open-source software designed to simplify how models run in any framework and is compatible with several GPU or CPU models for all inference types.
Following the beginning of the collaboration, Siemens Energy will run Nvidia Triton on AWS for scale and multi-tenancy, with plans to run at the edge where data cannot be moved out of the powerplants’ physical locations.
The company will also use Triton to monitor audio and video feeds for both efficiency and safety, and to perform additional pre-processing of the data, including person anonymization, of the images.
Lanner joins Nvidia GTC event
The network hardware provider participated in the virtual conference which took place last week, highlighting the benefits of OpenRAN (Radio Access Network) infrastructures for 5G applications.
Specifically, Sven Freundenfeld, CTO of Telecom Applications BU at Lanner, discussed the company’s collaboration with Nvidia in developing a scalable, hyper-converged edge platform for OpenRAN infrastructure.
The solution is based on Nvidia’s BlueField-2 DPU and features a programmable switching option capable of offloading workloads from open architecture CPUs.
Elma releases rugged AI platform based on Nvidia Jetson TX2i
Elma said it had developed a new solution for IVA, virtual reality (VR), augmented reality (AR), and AI applications at the edge.
Dubbed Jetsys-5320, the rugged platform offers an IP67 rating and is designed to increase visual intelligence in both transportation and defense applications.
These include high-resolution sensor systems, automatic target recognition, movement tracking security systems, and threat location detection and prediction.
Jetsys-5320 is powered by Nvidia’s Jetson TX2i System-on-Module (SoM), which enables real-time AI inferencing and DL/ML capabilities.
In addition, the module features dual-core Denver 2 64-bit CPU and quad-core ARM A57 Complex and 256 CUDA cores, Nvidia’s Pascal architecture capable of delivering 1.3 TFLOP of performance.
It also offers a wide range of connectivity options, supporting HD-SDI, Gigabit Ethernet with Power-over-Ethernet, USB 3.0 interfaces for video capture, and Mini PCIe expansion slots.
Article Topics
AIOps | data management | edge AI | Elma | Lanner | Mavenir | Nvidia | Siemens Energy
Comments