Axiomtek Reveals Edge AI Systems With NVIDIA Jetson Orin Nano and Orin NX Super Mode

[Nomad76]

Axiomtek - a world-renowned leader relentlessly devoted to the research, development, and manufacture of innovative, highly efficient, and reliable industrial computer products - is thrilled to unveil an electrifying upgrade that its AIE100, AIE110, and AIE900 series edge AI systems are going to gear up to support NVIDIA's game-changing Super Mode for NVIDIA Jetson Orin Nano and Orin NX series modules, launching this year with the NVIDIA JetPack 6.2 software development kit release.

This enhancement catapults performance to new heights, supercharging groundbreaking applications like local AI chatbots, real-time computer vision, and advanced robotics. Axiomtek's commitment to pioneering edge AI solutions just got even stronger, marking another milestone in its strategic collaboration with NVIDIA, as a Preferred member of the NVIDIA Partner Network with an Embedded Compute competency.



The introduction of Super Mode for Jetson Orin Nano modules unleashes powerful generative AI capabilities at the edge, delivering a 2x boost in generative AI inference performance and an astonishing 67 TOPS, paired with a 50% increase in memory bandwidth to 102 GB/s. From crafting LLM chatbots with retrieval-augmented generation to deploying cutting-edge AI-based robots, Super Mode for Jetson Orin Nano is set to redefine possibilities. The performance enhancements extend to the Jetson Orin NX modules, delivering an impressive 157 TOPS of AI performance, empowering developers with a seamless boost in generative AI capabilities.

Developers using Super Mode for the Jetson Orin Nano and Jetson Orin NX modules can also tap into NVIDIA's extensive AI software ecosystem, including NVIDIA Isaac for robotics, NVIDIA Metropolis for vision AI, and NVIDIA Holoscan for sensor processing. In addition, developers can harness the NVIDIA Omniverse platform for synthetic data generation and the NVIDIA TAO Toolkit for fine-tuning pre-trained AI models to accelerate their AI development and deployment processes.

"Axiomtek's AIE100, AIE110, and AIE900 series, powered by NVIDIA Jetson Orin Nano and Orin NX modules, will fully embrace the enhanced Super Mode capabilities," shared Annie Fu, Product Manager of Edge AI Solutions at Axiomtek. "Our edge AI systems are crafted to deliver superior computing power for diverse applications, from smart cities and industrial automation to robotics and AIoT. With NVIDIA's Orin Super Mode, our customers can harness the full potential of generative AI at the edge, leveraging cutting-edge tools like LLM, VLM, Vision Transformers, and Ollama. As a long-standing partner of NVIDIA, we're proud to integrate their state-of-the-art technologies into solutions designed for the future of generative AI workloads. This collaboration underscores our unwavering dedication to empowering customers with the advanced AI tools they need to tackle the evolving demands of edge applications."

Advanced Features of AIE Series

• Support Super Mode for NVIDIA Jetson Orin Nano and Orin NX modules (up to 157 TOPS)
• Ideal for local AI chatbots, real-time computer vision, and advanced robotics
• Diverse sensor and camera connectivity: USB, PoE, GMSL camera
• Seamless speed: 5G, Wi-Fi, LTE
• Supports 24/7 secure remote monitoring, control, and OTA deployment empowered by Allxon

View at TechPowerUp Main Site | Source

 

[Daven]

Orin seems to be everywhere. Why now? AI?

 

[silentbogo]

Orin seems to be everywhere. Why now? AI?

There are several reasons:
1) It's the most powerful "affordable" Nvidia Embedded SoC to date (which has been released only last year and not half-a-decade ago).
2) Significant price drop comparing to initial launch in 2022. It's half the price of the old Orin Nano, and AGX Orin is nearly 8 times more expensive and targeted towards different segment.
3) Compute modules ready right from the start, and not few years down the road
4) Excluding automotive stuff, Nvidia hasn't touched embedded segment seriously since the original Jettson Nano, which is still made and still running an 8+y.o. chipset (and still as hard to come by outside US as an albino unicorn born on the full moon). When Tegra X1 launched - there was just about as much hype, only in robotics and automation segment (also AI, just wasn't as accented at the time). All that's changed is a set of buzzwords used to market it. Everything in-between was too expensive and too "automotive".

 

[igormp]

since the original Jettson Nano, which is still made and still running an 8+y.o. chipset

The original one is not supported anymore software-wise, fwiw. If it's still being made, it's likely to comply with service agreement contracts.