NVIDIA has just announced a compact and affordable supercomputer, the Jetson Orin Nano Super Developer Kit, which would take the democratisation of generative AI capabilities among developers to a whole new level and that hobbyists or students. It introduces significant performance improvements over the previous generation that will democratise advanced AI development.
The Jetson Orin Nano Super is said to raise generative AI inference performance by up to 1.7 times, reaching 67 INT8 TOPS. It also boosts memory bandwidth by 50% to 102GB/s from its predecessor. These improvements will allow developers to run popular AI models more efficiently and enable the building of LLM chatbots, visual AI agents, and AI-driven robots.
Despite its small form factor, the Jetson Orin Nano Super still has enough juice to run multiple concurrent AI application pipelines. It has a 6-core Arm CPU and an NVIDIA Ampere architecture GPU with tensor cores, which makes high-performance inference possible. It supports up to four cameras, each offering much higher resolution and frame rates than earlier versions, perfect for robotics and computer vision projects.
The Jetson Orin Nano Super supports the full range of NVIDIA software, including Isaac for robotics, Metropolis for vision AI, and Holoscan for sensor processing. By extension, developers can tap into the larger Jetson community and open-source resources to ensure comprehensive support for their projects.
These upgrade path software updates will enable existing customers using Jetson Orin Nano Developer Kits to improve the performance in generative AI by a full factor of 1.7 over what is currently available, using additional new hardware investment.
The NVIDIA Jetson Orin Nano Super Developer Kit is an affordable, powerful, compact solution for one interested in developing skills in generative AI, robotics, or computer vision. It is very apt for transforming an innovative idea into reality because of enhanced performance combined with strong software support.