NVIDIA released the Jetson Orin Nano Super Developer Kit today. The new software release increases generative artificial intelligence workload performance from 40 trillion operations per second, or TOPS, to 67 TOPS and memory bandwidth from 68 GB/s to 102 GB/s. The result is up to 1.7X generative AI model performance gains and only 25 watts of power consumption.
The Nano is the smallest computer produced by NVIDIA. This edge computer is suitable for battery-powered mobile robotics, and it’s now smaller and cheaper than ever before.
NVIDIA said it continues to push the performance envelope for edge computing, enabling robotics developers to make their systems smarter and more capable.
Jetsen Orin Nano Super provides performance improvements over the baseline from the prior generation. | Credit: NVIDIA
Jetson Orin Nano Super can run LLMs, VLMs
The developer kit consists of a Jetson Orin Nano 8GB system-on-module (SoM) and a reference carrier board for prototyping edge AI applications. The SoM features an NVIDIA Ampere architecture graphics processing unit (GPU) with tensor cores and a six-core Arm CPU.
NVIDIA said this architecture facilitates multiple concurrent AI application pipelines and high-performance inference. It can support up to four cameras for higher resolution and frame rates than previous versions.
Nano Super can run almost any generative AI model, from large language models (LLMs) and vision language models (VLMs) to vision transformers. The best news for existing Orin developers is that existing customers can also increase their performance through a software update available on the NVIDIA website or the GitHub repo.
The Orin Nano 4GB Developer Kit also has a price break, with a new list price of $249, down from $499. With the price break and performance improvements, the result is a 3.5x greater value, said NVIDIA.
Last month, Rockwell Automation added NVIDIA Omniverse to its digital twin software. Rockwell Automation’s Emulate3D software now uses NVIDIA Omniverse application programming interfaces (APIs) to create factory-scale dynamic digital twins based on OpenUSD interoperability and NVIDIA RTX rendering technologies.
Also in November, NVIDIA added open AI and simulation tools for robot learning to accelerate the development of robots including humanoids.
Can I run Ominiverse in this without an rtx in my laptop
Over a month later and I can’t find a single place that has stock. Any news on this?