Frore Systems Introduces AirJet Cooling Technology for NVIDIA Jetson Orin Nano Mini-Supercomputer

Frore Systems has recently introduced an innovative cooling solution specifically designed for NVIDIA’s Jetson Orin Nano Super AI module. This development promises to enhance performance capabilities significantly.

Enhancing NVIDIA’s Jetson Performance with Frore Systems’ Advanced Cooling Technology

[Press Release]: The newly unveiled Jetson Orin Nano Super, a compact yet powerful AI “mini-supercomputer,”boasts an impressive capacity of executing 67 trillion operations per second (TOPS) while consuming only 25 Watts. However, such expansive AI capabilities generate substantial heat, which, without efficient cooling, may lead to performance throttling.

This high-performance architecture allows for the execution of advanced AI models across various applications, including NVIDIA Isaac for robotics, NVIDIA Metropolis for vision AI, and the NVIDIA Holoscan for sensor processing. Moreover, tools such as NVIDIA Omniverse Replicator for synthetic data generation and the TAO Toolkit for model fine-tuning can function effectively on the Edge. These innovations promote enhanced compute efficiency, reduced latency, and increased data security.

The necessity for an effective heat dissipation solution cannot be overstated. Without it, the Jetson Orin Nano Super risks throttling its performance, thereby limiting the potential of Edge AI applications across diverse sectors, including robotics, industrial automation, smart cities, healthcare, and retail analytics. Frore Systems’ AirJet PAK 5C-25 delivers an unparalleled cooling solution, ensuring the Jetson module achieves its full 25 Watts performance potential.

The AirJet PAK 5C-25 is a state-of-the-art, self-contained active cooling module that is not only thin and quiet but also dustproof and water-resistant. Designed to maintain peak performance in challenging environments, it effectively removes up to 25 Watts of heat. This compact solution supports the Jetson Orin Nano Super even within industrial-grade enclosures that are ultra-compact, silent, and vibration-free.

When compared to traditional fan cooling solutions—which can be bulky, noisy, and susceptible to dust and moisture ingress—the AirJet PAK is significantly smaller and more efficient. In fact, it is 60% more compact than fan-based options, making it an ideal choice for diverse AI application scenarios.

As the first of its kind, the AirJet PAK is engineered to enhance a variety of System on Module (SoM) AI computers. This includes compatibility with NVIDIA’s Jetson Orin Nano, Nano Super, NX & Orin AGX modules, as well as SoMs from prominent manufacturers like Qualcomm, NXP, and AMD/Xilinx. The modular design facilitates direct mounting on the SoM, optimizing temperature management and maximizing Edge AI performance.

AirJet PAKs come in various sizes for seamless integration into Edge AI platforms tailored to differing performance needs.
AirJet PAK 5C-25: Features 5 AirJet chips, dimensions of 100x65x9.8mm, able to dissipate 25 Watts of heat, supporting up to 100 TOPS.
AirJet PAK 3C-15: Contains 3 AirJet chips, measuring 100x65x5.8mm, removing 15 Watts of heat with support for up to 40 TOPS.
AirJet PAK 1C-5: Equipped with 1 AirJet chip, dimensions of 30x65x5mm, effective heat removal of 5 Watts, supporting up to 10 TOPS.

All AirJet products offer scalability; multiple AirJet PAKs can be combined to address higher performance and heat dissipation requirements. For instance, two AirJet®PAK 5C-25 units can collectively remove up to 50 Watts, supporting processing demands of up to 200 TOPS.

Frore Systems will showcase this innovative technology at CES in January 2025, providing demonstrations of Industrial Edge AI platforms utilizing AirJet PAK, along with consumer electronics enhanced by AirJet for superior performance. Featured products include the Samsung Galaxy Book4 Edge, MacBook Air, and iPad Pro, as well as numerous commercial offerings already on the market. Frore Systems continues to push boundaries in performance for Edge AI devices.

Source & Images